MACINTYRE-Against The Self-Images of The Age
MACINTYRE-Against The Self-Images of The Age
MACINTYRE-Against The Self-Images of The Age
This ebook has been made possible in part by a major grant from the
National Endowment for the Humanities: Exploring the human
endeavor. Any views, findings, conclusions, or recommendations
expressed in this book do not necessarily represent those of the National
Endowment for the Humanities.
Introduction vii
PART ONE
PART TWO
Hinden (London: Allen & Unwin; New York: Pantheon Books, 1964).
38
The socialism of R. H. Tawney 39
democracy, accompanied by a preface by Rita Hinden and by
Gaitskell's address at the 1962 Memorial Service for Tawney,
makes an illuminating book.
The heart of the matter for Tawney is the moral deficiency of
capitalism.
The revolt of ordinary men against Capitalism has had its
source neither in its obvious deficiencies as an economic engine,
nor in the conviction that it represents a stage in social evolu-
tion now outgrown, but in the straightforward hatred of a
system which stunts personality and corrupts human relations
by permitting the use of man by man as an instrument of
pecuniary gain. . .. "
It is this demon-the idolatry of money and success-with
whom, not in one sphere alone but in all, including our own
hearts and minds, Socialists have to grapple.
Sentences like these are scattered throughout Tawney's writings.
One need not be a cynic nor an immoralist to find so much cliche-
ridden high-mindedness suspect. The answer of his admirers
may be to stress, as Gaitskell does in his address, Tawney's
personal goodness-"! think he was the best man I have ever
known." The difficulty is that what both the reminiscences and
Tawney's own writings communicate is a banal earnestness rather
than the manifold virtues ascribed and praised. It is fairly clear
what is missing. The moral denunciation of British capitalism
took its content and its interest not from the morality of socialists
but from the immorality and evil of capitalism. What we miss in
these essays is the social context of the 1920s, of poverty, of un-
employment, of suffering.
Moreover the immediacy of these evils was linked with a hard-
headed, common-sense practicality about their cure. Public
ownership of the coal mines or the railways in Britain was not a
radical solution; that it was the only solution, was implicitly
acknowledged by the lack of Conservative opposition when the
measures were finally put through Parliament in the late 1940s.
But why did it take so long to achieve this solution? A govern-
ment commission headed by Mr. Justice Sankey and including,
along with Tawney, men of widely different views had recom-
mended the nationalization of the mines in 1919. The reason for
the delay lies in the failure of nerve in Britain's ruling class be-
tween the two wars. The politicians of the age-MacDonald,
Snowden, Bonar Law, Baldwin, and Chamberlain-are in per-
spective tiny and impotent figures. No wonder that in comparison
Agat'nst the Self-Images of the Age
with them Tawney assumed the appearance of great moral stature.
Yet if he appears impressive by contrast, we must also ask whether
in many ways he did not share many of the attitudes and indeed
illusions of his contemporaries.
Tawney equated capitalism with private capitalism, and private
capitalism with the effective sovereignty of the functionless share-
holder. He defined socialism on at least two levels, both of which
were inadequate. At one level he meant the moral values of
fraternity and equality, which are, unhappily, terms too vague
and general for political guidance until they are embodied in
specific social practices and institutions. At another level he
defined socialism by his concept of capitalism: the replacement
of private ownership by public ownership or control and the
state's acceptance of responsibility for social welfare. Thus he
never took stock of the capitalism of the big corporation-the
capitalism which may for its own purposes accept trade unionism,
the welfare state, and even measures of state intervention and
public ownership. He is in fact oblivious not merely of Keynes,
but of the kind of capitalist ethos in which neo-Keynsian politics
could be made effective.
Yet is it not perhaps absurd to criticize Tawney for being
limited by the horizon of his period? Not if what we are criticizing
is above all lack of political intelligence and imagination. The
lack of political imagination is notably present in his estimate
(reprinted in The Radical Tradition) of the role and achievement of
the postwar Labour government. He profoundly underestimates
the continuity of that government with the wartime coalition
government. He writes of the Labour ministers as if they were by
deliberate choice implementing socialist policies, when in fact
they were providing the necessary and inevitable solutions to the
problem of laying a new basis for British capitalism. He never
mentions the frustration and disillusionment that that govern-
ment engendered, especially among its working-class supporters.
To say this is not to underrate the achievements involved in im-
plementing the 1944 Education Act (passing it was the work of
the wartime coalition government), or of the handing over of
power in India, or of the creation of the National Health Service.
It is to say that any intelligent pragmatist, thoroughly but far-
sightedly imbued with capitalist values, could not and would not
have done otherwise. And it is not only that Tawney underrated
the resources of an intelligent conservative defense of capitalism.
In his statement of socialist objectives he is curiously blind to
how greatly his declared ends and his chosen means were at odds
The socialism of R.H. Tawney 41
with one another. He cared passionately that workers should
extend their control over the work process; and he wanted,
probably more than anything else, to democratize the British
educational system. Yet the kind of orthodox Labour Party politics
in which he put his hope has always been managerial and merito-
cratic. The Labour Party has shown immense hostility to those
rank-and-file trade unionists who have been concerned with
issues of workers' control; and it has shown a simple lack of
interest in many less radical measures concerned with democracy
in industry. In education the Labour Party's support for compre-
hensive schools and for equality of opportunity did not, when it
was in office, prevent it from helping to create through the 1944
Education Act a class system in education which not only favors
the middle-class child, but has helped to create new class barriers.
Labour is increasingly the political expression not of workers, but
of managers and technocrats. It is the party of the other half of
our ruling class.
Why did Tawney succeed in concealing from himself as well as
from others the extent to which the British Labour Party is
merely an alternative Conservative Party? One answer can be found
in The Radical Tradition. Tawney did not lack that essentially
English quality, insularity. It is no accident that there is little in
his book about peace or international socialism. In his essay on
"Social Democracy in Britain" he asserts that "it is not for a
foreigner to discuss" the standing of capitalism in the United
States. And he appears to restrict himself not only geographically
but theoretically. We have jibes-not arguments-against Marxism,
and economic expertise is treated as a topic for a joke. The limits
of theoretical inquiry appear to be those which actually exist in
the House of Commons, a not very theoretical body.
Tawney thus appears to define politics itself as what might go
on in a British Parliament. Since the role of Parliament, and con-
sequently of electoral politics, in the decision-making processes of
British life has steadily declined, it is not surprising that already
his writings have a curiously antique air. He never even asks
whether Parliament may not be among the institutions which
need democratizing. And however radical he may be about the
economic activities of private capitalism, he is a true member of
the Labour Party in being completley complacent about British
political institutions.
So a book of essays designed to celebrate "the Democratic
Socialist philosopher par excellence" is in fact a monument to the
impotence of ideals. It is not that Tawney failed to live up to his
42 Against the Self-Images of the Age
ideals or to propagate them. He succeeded admirably. Nor is it
that his ideals were insufficientlyhigh. It is simply that the Socratic
question of whether one would rather have one's shoes mended
by a good cobbler or a good man has relevance in politics too.
Goodness is not enough.
5
How not to write about Lenin 1
Of the books below2 the two most important are Gerassi's collection
of Che Guevara's writings and Rojo's brilliant biography. Guevara's
own reminiscences of the Cuban revolutionary war are interesting,
but the reader needs to be well informed already to make much
use of them. Debray's theorizing is perhaps only interesting for the
contrast between the Debray version of Che and Che as he was,
and the Sartre is worth noticing in this context because it helps us
to judge how much of Debray is Paris academicism. Finally I
notice the American version of Che's diaries merely to note that
it differs in important ways from the Cuban version. The publishers
on their dust jacket say their edition "was authenticated not by
Cubans or Bolivians but by Americans"; here's news for you,
Stein and Day-I still do not trust it. James accuses Che of
"personal pique" ; his publishers join with him in entertaining
the suggestion that Castro was jealous of Che, deliberately denied
the help he could have given, and so betrayed him. This obscene
suggestion does not come well from Americans, who ought at
least to realize that the death of Guevara may well cost them as
much as his life did.
The death of Che had-and it is difficult to use the word after it
has been so cheaply misused-tragic quality. To use a dramatic
metaphor is not to suggest anything histrionic about Che's actions
1 Reprinted from Partisan Review, 1969.
2 A review of John Gerassi, ed., V enceremos! The Speeches and Writings
of Che Guevara (New York: Macmillan; London: Weidenfeld & Nicolson,
1969; Che Guevara, Reminiscences of the Cuban Revolutionary War (New
York: Grove Press; London: Allen & Unwin, 1968); Daniel James, ed.,
The Complete Bolivian Diaries of Che Guevara (New York: Stein & Day;
London: Allen & Unwin, 1969); Ricardo Rojo, My Friend Che (New
York: Dial Press, 1969); Regis Debray, Revolution in the Revolution?
(New York and London: Monthly Review Press, 1967); Jean-Paul
Sartre, The Communists and Peace (New York: Braziller, 1968).
70
Marxism of the will 71
or passions; it is to indicate that those actions and passions are an
appropriate subject for poetry as well as for history, because, as
Aristotle said, poetry is "more universal" than history. Che was
not just an individual, but a representative figure, who lived out
a tragic action. A tragic action is one in which a hero encounters a
catastrophe as a result of a flaw in his character. By character I
do not mean a mere assemblage of psychological traits, I mean
rather the incarnation of a role. (What poetry was for Aristotle,
sociology is for us.) What was Che's flaw?
To ask this question, I have suggested, is to ask about a role
and not about an assemblage of personal traits. That personal
traits can explain little in political or social action is made clear
once again in reading reminiscences of Che by those who knew him
well. He was an asthmatic who developed a will strong enough to
take him onto the athletic field and through medical school. He
was an ascetic who did not undervalue sex or alcohol. He was an
altruist, but without any signs of that self-contempt which so
often underpins altruism. I shall suggest later that these traits
were not entirely unimportant in relation to some key positions
that Che took up; but there are no splendid psychological generali-
zations to be constructed which will demonstrate that asthmatic,
ascetic altruism is the seedbed of revolution. As so often, what is
impressive is not the connection, but the relative lack of connec-
tion between individual personality and social role. The need to
reminisce about Che has in any case obviously little to do with any
task of explanation; it is much more as though his friends still
have to reassure themselves that it all really did happen, that this
living out of one of our political dreams was not in fact only a
dream.
The search for such reassurance is perhaps connected with the
extent to which the Cuban Revolution was an accidental happen-
ing. By this I mean much more than that it did not follow out the
patterns of previous revolutions. Regis Debray is able to emphasize
that and yet to insist that the Cuban Revolution embodied an
experience from which more generally applicable laws and maxims
can be extracted. In this he follows Che faithfully and yet there
is an important difference in tone between what Huberman and
Sweezy call Debray's "comprehensive and authoritative presenta-
tion of the revolutionary thought of Fidel Castro and Che Gue-
vara" and what we actually encounter in Che's writings. This
difference arises from the stale, academic atmosphere of Debray's
arguments. For however authentically Debray may reproduce
what is new and Cuban, he does so in a setting and a style which
Against the Selj-11llllgesof the Age
is old and French. So that while in Guevara's own narratives the
Marxism-Leninism somehow coexists with a sense of the Cuban
Revolution as a chain of improvisations and coincidences, in
Debray's writing revolutionary action becomes nothing but matter
for theoretical formulas counterposed to other theoretical formulas.
Accident has disappeared and with it truth.
We find in Debray a constant reiteration of Sartrian themes.
There are the same strange attempts to unite historical necessity
and absolute freedom, to dissent from Stalinism and yet to count
Stalin among the revolutionary ancestors, and to portray Trot-
skyism as the villain of the piece; indeed Debray explicitly refers
back to Sartre's fifteen-year-old anti-Trotskyist polemic in
The Communistsand Peace.One can well understand why Trotsky's
ghost haunts Sartre and Debray. For both Sartre and Debray
have a peculiar conception-far more elitist than that of Leninism
--of an inert mass of be it workers, be it peasants, who need a
leadership of particular gifts to rouse them to revolutionary
activity. Sartre in 1952 and 1954 was equally contemptuous of
those sociologists who declared that the French working classes
were not revolutionary and those Trotskyists who declared that
they were revolutionary-but that their revolutionary tendencies
were suppressed or inhibited by their reaction to the Stalinist
leadership of the Communist Party of France. In Sartre's view
the working masses are not, but will become, a revolutionary
class precisely because the Communist Party presents them with
goals which transcend their immediate needs; so for Debray the
guerrilla army is to present the peasants with goals which trans-
cend their immediate needs. It is a doctrine which enables Sartre
and Debray to set on one side in the most arbitrary way the
question of what workers or peasants do in fact want now. It also
enables Sartre to disregard the theoretical positions of St~linist
bureaucrats; his understanding of the falsity of Stalinism seems
in his writings of the early 1950s only marginal to his evaluation
of Stalinism's political function. So Debray too exalts questions of
organization over questions of political goals and programs and
sneers at the Trotskyists for their emphasis upon fundamental
theory.
In his intellectual style then Debray is unlike Che; but Che
himself could not avoid facing dilemmas which in other contexts
were responsible for creating Trotskyism, and he could not avoid
making choices which were incompatible with Trotskyism. This is
because Trotsky himself had had to face at successive points in his
career all the dilemmas of those who wish to make a Marxist
Marxism of the will 73
revolution in an underdeveloped country and because too the
failure of Trotskyism to provide a recipe for successful revolu-
tionary practice in the face of those dilemmas is an inescapable
fact. What is the part of the peasantry in the making of a Soci.alist
revolution? Marx could see no part for them, Mao invented one
ex nihilo and called it Marxist, and every position intermediate
between Marx's and Mao's has been taken up by some Marxist
theorist at some time. Trotskyism at the very least represents the
thesis of the ineliminable necessity of the participation of an
industrial working class in revolution-making. Can there be
soci.alismin one country? One paradox of post-Stalin Stalinism is
that it may be those who are most repelled by the surviving
Stalinist features of the Soviet Union who therefore try to build a
socialist revolution in isolation from the Soviet camp or at least in
the minimum of contact with it. But in so doing they revive the
very thesis of "socialism in one country" on which Stalinism was
founded and in this way reject Trotskyism. What is the place of
the revolutionary party? The orthodox Communist Parties in
Latin America are obviously not revolutionary parties; their weak-
ness and their reformism are notorious. But in the struggle waged
by peasant guerrillas there is little room for a party at all. Hence
Trotskyism once more appears as the ghost of orthodox Bolshev-
ism, repudiating militantly the only militant strategies apparently
open in Latin America.
Guevara's position is thus easily defined by contrast with that
of Trotskyism, and in this at least Debray is perceptive. But if
Guevara offered us a revolution made by peasants, a revolution
which creates socialism in one country, and a revolution with a
revolutionary army rather than a revolutionary party, he aspired
to do so as a Marxist-Leninist, and here is the crux. For if Bol-
shevism can only appear in the modern world in ghostly form,
Trotskyism is indeed its authentic ghost. How then can an anti-
Trotskyist position be grafted on to Marxism-Leninism? To answer
this question will return us to my initial inquiry as to the tragic
flaw in the role acted out by Che. For what Che uses to close the
gap between what the Marxist-Leninist must hold on an objective
analysis to be a situation in which the socialist revolution cannot
yet be made and the revolutionary aspirations of the selfsame
Marxist-Leninist who confronts himself with this, as it must
seem, defeatist analysis, is an appeal to pure will. Lenin too was
confronted with this gap and at every stage wrestled to link the
present and the future by means of a consciousness nurtured by
the organizational forms of the party. In Guevara, although
74 Against the Self-Images of the Age
questions of organization are treated with intellectual respect, it
is the voluntarist component of Leninism which is appealed to as
never before.
Consider for example the question of planning. Guevara con-
ducted a polemic against the French Communist expert on
planification, Charles Bettelheim, in which he argued that because
of the level that consciousness (in the Marxist sense) had reached
in the world at large, the social and political consciousness of
Marxists in a country where the objective conditions for a socialist
revolution had not yet been reached could none the less enable
them to transcend those limitations and to do what seemed
objectively impossible. From this premise Che argued further
in more general terms for a relative independence of cultural
superstructure from economic basis. This led him to quarrel
with Bettelheim and other Marxists on economic policy. Material
incentives, such as may be provided by a wages structure may be
appropriate as the mainspring of a market economy, but are
inappropriate to socialism. Centralized planning demands the
centralization of major economic decision-making, but it does not
require centralized management.
What is to take the place of material incentives and of the
dictates of centralized management? A new motivation springing
from the new nature of socialist man. Moral incentives must be
the mainspring; material incentives must be subordinate. The
word "moral" recurs throughout Che's writings. He was the
minister who awarded the title "Hero of Labor" to workers who
excelled. In his speeches to workers he constantly urged sacrifice
and hard work. His personal asceticism put his right to make such
calls beyond question. But their theoretical justification is quite
another matter.
Behind the Leninist voluntarism we see in Che the revival of an
older answer to the Marxist dilemma about morality. Marx him-
self never raises explicitly the question of the motives of those who
seek to achieve socialism. At the turn of the century when Bern-
stein raised the question of the moral foundations of socialism
and turned back to Kant's invocation of duty in order to answer it.
Kautsky replied to him with a crude invocation of utilitarianism
which relied on an underlying appeal to material self-interest;
and Rosa Luxemburg in her polemics against Bernstein avoided
coming to grips with this question at all. Bernstein's Kantian
answer was in fact more influential than we sometimes realize;
and to be Kantian was not necessarily to be a right-wing social
democrat. After 1914 Kautsky the orthodox Marxist was far to the
Marxism of the will 75
right of Karl Liebknecht, the Kantian and Spartacist. Guevara
was Karl Liebknecht's spiritual heir; like Liebknecht he in the end
bore witness to the fact that moral heroism is not enough. In the
improbable environment of Cuba, Kantian moral theory was
reborn as revolutionary.
Che's moral heroism, his attempt to transcend the material
environment, was the tragic flaw which finally destroyed him in
Bolivia. When he left Cuba he wrote to his children: "Above all,
always be able to feel deeply any injustice committed against
anyone in any part of the world. It's a revolutionary's most
beautiful quality"; and to his parents: "Essentially, nothing has
changed, except that I am much more conscientious, my Marxism
has struck deep roots and is purified." Again the Kantian note is
struck. Conscientiousness took him to his death, because it led
him to ignore political and military facts, and especially Barrientos'
ability to mobilize peasant support.
When I stress Che's moralism, I do not want to underestimate
his intellectual qualities. Americans in particular should read the
speech rejecting the Alliance for Progress made at the Punta del
Este Conference of the OAS in 1961; about that particular
Kennedy cloud-cuckoo project Guevara has proved alarmingly
right. But when Guevara is not being critical of imperialism, he is
all too apt to substitute invocations of honor or of the spirit of
sacrifice for intellectual analysis. Guevara's student admirers are
indeed moved precisely by this and so is John Gerassi who has
done scholarship a service by his collection of Guevara's speeches
and writings. Yet what they admire is just that abstract moralism
which Marx himself ought to have taught us to suspect. Che's
last letter to his parents begins with an allusion to Cervantes:
"Once more I feel Rocinante's ribs under my heels; I'm taking to
the road again with my shield on my arm." Perhaps as he wrote
this he should have remembered that other reminiscence of
Cervantes in a footnote in Capital which ends by Marx remarking
that "Don Quixote long ago paid the penalty for wrongly imagin-
ing that knight errantry was compatible with all economic forms
of society."
IO
death in 1970 robbed us of the finest and most intelligent Marxist of the age.
80 Against the Self-Images of the Age
Sometimes we get an acute criticism of Cartesianism of purely
philosophical interest:
If man were to begin by studying himself, he would see how
incapable he is of going beyond himself. How could it be
possible for a part to know the whole? But he may perhaps aspire
to a knowledge of at least those parts which are on the same
scale as himself. But the different parts of the world are all so
closely linked and related together that I hold it to be impossible
to know one without knowing the other and without knowing
the whole (fragment 72).
At other times we find the whole of human knowledge brought
under condemnation:
Everything here on earth is partly true and partly false. But
essential truth is not like this, for it is wholly pure and wholly
true. The mixture that we find here on earth both dishonours
and destroys this truth ... (fragment 385).
Is Pascal simply inconsistent? Should his solution have been in a
tough-minded way to grasp one of the horns of his dilemma and
abandon the other? This would have been the Cartesian solution
as it would also have been the Augustinian. But Pascal inhabits
two conceptual universes the claims of which he can neither
reconcile nor abandon. Torn as he is between two realms, he can
see each from the point of view of the other and his own predica-
ment from both. Thus from within Christianity he sees his dilemma
as itself prefigured by Christian theology. For does not Christian
theology assert that we inhabit two realms, that man belongs both
with the angels and the beasts, that if human nature ignores its
limitations and seeks to be angelic it becomes bestial, that a hidden
God has revealed himself incarnate and so on? The paradoxes of
Christianity show it to be divine.
Yet from within the world he can see Christianity in the per-
spective of his own critique of Cartesianism. His skepticism about
clear and distinct ideas ("Too much clarity darkens") and about
any allegedly indubitable first principles, even those of skepticism,
extends to any alleged arguments for Christianity, even his own.
The theory of chances, which he had elaborated to assist his friend
Mere at the gambling tables, encounters its limit at the point at
which there are no more probabilities, but the stakes are infinite.
Yet at this very point a wager cannot be avoided. It is only
through a wager that God exists that meaning is conferred on an
otherwise meaningless world. Yet it is from the standpoint of that
Pascaland Marx: on Lucien Goldmann'sHidden God 81
world that we have to learn that belief in God has to accept the
status of a wager.
Let Pascal abandon the world and he becomes the ancestor of
Kierkegaard, of a self-contained fideism. Let him abandon
Christianity and he becomes the ancestor of Hume, avoiding
skepticism only by calling nature and custom to his aid. His
greatness is in abandoning neither. Why? To understand Gold-
mann's answer to this question we must turn to his use of Marx
and Lukacs.
The danger is that we read what Goldmann has to say through
our own preconceptions; and where Marxism is concerned no one
is without preconceptions. Goldmann's thesis is that Pascal ex-
pressed in one particular form a coherent world vision which
Lukacs was to characterize. That world vision, the vision of
tragedy, is rooted in the social history of Jansenism, expressing
the attitudes implicit in the predicament of the noblessede robe.
Our preconceptions and prejudices might lead us to treat Gold-
mann's views as just one more explanation of the history of thought
in terms of an economic and social basis. But if we did we should
miss the concreteness of Goldmann's concerns. He is very far
from forcing the interpretation of J ansenism and Pascal into an
already existing theoretical structure. Rather it is at least partly
through his studies of J ansenism and Pascal that he gives meaning
to his theoretical terms. So one cannot fully understand the early
theoretical chapters of his book until one has read the later his-
torical and literary studies. Pascal and J ansenism are made to
illuminate Marxism quite as much as Marxism is made to illu-
minate Pascal and J ansenism. Pascal himself would have under-
stood this: "The last thing one discovers when writing a book is
what ought to have come first."
The tragic vision, which Lukacs described, 1 is the vision of a
world where God is no longer present, and yet even in his absence
life has to be lived out by the tragic hero with the eye of God upon
him. Because God is absent, the hero cannot succeed in the world.
Because God, though absent, still regards him, he cannot abandon
his task. He is the just man under condemnation, whom the critics
of J ansenism saw at the heart of Jansenist doctrine. So long as he
responds by refusing the world, he is Barcos. So long as he tries
to live in the world and yet also to refuse it, he is Pascal himself.
"In Die Seele und die Formen (Berlin: Essays Fleischel, 1911). The
Lukacs whom Goldmann follows is the since self-condemned Lukacs of
this book and of Geschichte und Klassenbewusstsein. Lukacs has now (1971)
altered his attitude to his work of this period yet once more.
82 Against the Self-Images of the Age
The Lukacs of 1911 saw the tragic vision as one form of aesthetic
insight; Goldmann sees it as.capable of embodiment only when it
expresses a form of social life which can recognize its own crisis
in this vision.
The noblessede robe (as contrasted with the noblessede cour) was
composed of those lawyers and administrators whom the French
monarchy used in achieving hegemony over the rest of the nobility,
strengthening itself by this alliance with the Third Estate and the
townsmen. During the seventeenth century the monarchy breaks
this alliance and becomes an independent power, balancing class
against class and governing through its own corpsde commissaires.
The noblessede robe thus find themselves on the defensive; their
allegiance to the crown and to the established order is the condition
of their flourishing and yet now the crown has less and less use for
them. They can less and less live out the only role they know, and
yet they must recognize the legitimate authority of the power that
is abolishing that role. The congruence of this social experience
with the tragic vision is clear. (One is reminded of Milton who does
not just have to justify the ways of God to man in general, but has
to reconcile the hidden fact that God rules with the manifest fact
that Charles II rules and the saints do not.) Thus the parlement's
manifest sympathy for J ansenism is for Goldmann a sign of
recognition by a segment of the middle class that in J ansenism
their own fragmentary attitudes receive completer expression and
endorsement than elsewhere.
Goldmann is at the opposite extreme from those self-styled
Marxists who have tried to reduce the artist or the philosopher
to a mere product of his social background. He sees that such a
reduction fails to account precisely for what interests us in a
writer's achievement, his distinctiveness. Goldmann's injunction
is rather that we should understand the background through the
writer, seeing in the coherence of great art or great philosophy
something that is only implicit in the thought and action of
ordinary men. So he invites us to understand J ansenism through
Pascal, and the noblessede robe through Jansenism. Moreover,
the greatest writers both express and transcend their age. They
show us the possibilities in the age of going beyond it, whereas
lesser writers exhibit the limitations imposed upon them by the
age.
It is not only Pascal whom Goldmann views in this light. He
analyses Racine's tragedies in terms of the concept of the tragic
vision, seeing a parallelism between the J ansenism of refusal of the
world and the tragedies of refusal, Andromaque and Britannicus,
Pascal and Marx: on Lucien Goldmann's Hidden God 83
while Pascal's attitude is paralleled in Phedre. These parallelisms
are brought out within a much more detailed classification of
tragedy. The justification of this classification and the use of
Lukacs' artificial construct of the tragic vision can lie solely in
whether it enables us to understand better not only the plays
themselves, but also the author's relationship to them. And
Goldmann follows Racine himself in seeing the heart of Jansenism
in Phedre. For it was in the preface to Phedre that Racine hoped
that his method in this play "would perhaps be a way of reconcil-
ing with tragedy a number of persons famous for both their piety
and doctrine"-although he does so ostensibly for the platitudi-
nous reason that in his play virtue and vice receive their deserts.
Whereas in fact if Goldmann is right the greatness of Phedre
herself is that she cannot refuse the claims of the world as em-
bodied in her own passion and her conception of Hippolytus but
nor can she refuse to live with the eye of God upon her. Of Phedre
what Lukacs wrote of the tragic hero holds:
He hopes that a judgment by God will illuminate the different
struggles which he sees in the world before him, and will reveal
the ultimate truth. But the world around him still follows the
same path, indifferent to both questions and answers. No word
comes from either created or natural things, and the race is not
to the swift nor battle to the strong. The clear voice of the
judgment of God no longer sounds out above the march of
human destiny, for the voice which once gave life to all has now
fallen silent. Man must live alone and by himself. The voice of
the judge has fallen silent forever, and this is why men will
always be vanquished, doomed to destruction in victory even
more than in defeat.
Tragic thought is not simply an episode in the past. Pascal, in
Goldmann's view, is not only illuminated by Marx and Lukacs, he
is their ancestor. He anticpates their epistemology in two crucial
respects. First of all, he understands that the knowledge of man
himself depends on grasping the individual as part of a totality.
Yet we cannot grasp the totality except insofar as we understand
the individuals who comprise it. Marx wrote:
A loom is a machine used for weaving. It is only under certain
conditions that it becomes capital; isolated from these condi-
tions it is as far from being capital as gold, in its natural state,
is from being coin of the realm.
What are these conditions? They include both the existence of a
Against the Self-Images of the Age
whole system of economic activity and the informing of human
activities and intentions by concepts which express the relation-
ships characteristic of the system. We identify a loom as capital or
gold as coin only when we have grasped a whole system of activities
as a capitalist or monetary system. The individual object or action
is identifiable only in the context of the totality; the totality is only
identifiable as a set of relationships between individuals. Hence
we must move from parts to whole and back from whole to parts.
Goethe, Hegel, and Marx all grasped versions of this truth about
the human sciences. Pascal, as Goldmann interprets!him, uses it
against Descartes in the fragments about the whole and the parts
and about figure and motion, which I quoted earlier. We can put
the essence of his criticism by saying that, just as no amount of
mechanical explanation of the working of a loom will tell us what
weaving is, or how a loom becomes capital, so no amount of
mechanical explanation of reflexes will tell us what human action
is or how a man becomes in his actions like an angel or a beast.
For that we need to understand human action as part of a total
system in which certain norms are established. The difficulty is
that men have false as well as true consciousness of the systems
of which they form a part. They need a criterion for discriminat-
ing true from false, and they exhibit this need especially in trying
to understand the over-all context of their actions. For Pascal this
context is provided by God and his will; for Hegel and Marx by
the history of society. For Pascal the contradictions involved in
the task are ultimate and irresoluble; for Hegel and Marx they
can be transcended in a future form of human community. But
if tragic thought and dialectical thought differ in these crucial
respects, they also resemble each other at key points. Both know
that one cannot first understand the world and only then act in it.
How one understands the world will depend in part on the deci-
sion implicit in one's already taken actions. The wager of action is
unavoidable. Goldmann is willing even to use the word "faith" of
the Marxist attitude, and he sees a real continuity between Augusti-
nian theology and Marxism, despite their differences on such
issues as the actual existence of God:
There are only three kinds of person: those who, having found
God, seek Him; those who, not having found Him, spend their
time seeking Him; and those who live without having found
Him and without seeking for Him either. The first are both
I
Axe moral judgments essentially and necessarily universalizable ?
The contention that they are is expressed in its most illuminating
form in R. M. Hare's paper on "Universalizability." 2 Hare
borrows his terminology from E. Gellner's paper on "Logic and
Ethics," 3 where Gellner distinguishes what he calls U-type and
E-type valuations. A U-type valuation is an application of "a
rule wholly devoid of any personal reference, a rule containing
merely predicates (descriptions) and logical terms. " 4 An E-type
valuation is one containing some uneliminable personal reference.
Hare's thesis is that moral judgments are U-type valuations.
To give a reason for an action is not necessarily to commit oneself
to such a valuation "for I see no grounds in common language for
1 Reprinted from Philosophy, 1957.
1 Proceedings of the Aristotelian Society (1954-55), pp. 295-312.
3 Ibid., pp. 157--78.
'Ibid., p. 163.
96
What morality is not 97
confining the word 'reason' to reasons involving U-type rules. " 1
But Hare goes on to say that his thesis "is analytic in virtue of the
meaning of the word 'moral.' "
What this amounts to is made very plain in an imaginary
conversation which Hare constructs between a "Kantian" and an
"Existentialist." This runs as follows:
E. : You oughtn't to do that.
K. : So you think that one oughtn't to do that kind of thing?
E. : I think nothing of the kind; I say only that you oughtn't
to do that.
K.: Don't you even imply that a person like me in circum-
stances of this kind oughtn't to do that kind of thing when the
other people involved are the sort of people that they are?
E.: No; I say only that you oughtn't to do that.
K. : Axe you making a moral judgment?
E.:Yes.
K. : In that case I fail to understand your use of the word
"moral."
Hare's comment on this is: "Most of us would be as baffled as
the 'Kantian'; and indeed we should be hard put to it to think of
any use of the word 'ought,' moral or nonmoral, in which the
'Existentialist's' remarks, would be comprehensible. Had the
'Existentialist' said 'Don't do that,' instead of 'You oughtn't to
do that,' the objections of the 'Kantian' could not have been made;
this illustrates one of the main differences between 'ought' and
ordinary imperatives." 2
The crux then of Hare's position is the contention that when-
ever anyone says "I, you or he ought to do so-and-so," they are
thereby committing themselves to the maxim "One ought to do
so-and-so." This commitment is embodied in the meaning of the
word "ought" insofar as "ought" is used morally-and indeed,
Hare seems to say, in nonmoral uses of "ought" also. But is this
contention in fact correct? Consider the following example which
is borrowed from Sartre.3 One of Sartre's pupils was confronted
during the war with the alternatives of leaving France to join de
Gaulle or of staying to look after his mother. His brother had been
killed in the German offensive in 1940 and his father was a
collaborator. These circumstances had left him with a strong
feeling that he was responsible as a patriot and they had left his
1 Ibid., p. 278.
1 Ibid., pp. 304-5.
• L'Existentialisme est un Humanisme, pp. 3g-42.
98 Against the Self-Images of the Age
mother in a state of almost complete dependence upon him. What
should he do? Stay with his mother or escape to England? Sartre
uses this problem in order to argue that there are no "objective"
criteria by which such a choice may be made. Part of the force of
his argument is this. Someone faced with such a decision might
choose either to stay or to go without attempting to legislate for
anyone else in a similar position. He might decide what to do
without being willing to allow that anyone else who chose
differently was blameworthy. He might legitimately announce his
choice by saying, "I have decided that I ought to stay with my
mother." If he did so, his use of "ought" would not express any
appeal to a universalizable principle. It would not be a U-type
valuation, but it would be a moral valuation.
Two points need to be made about this example, The first
concerns the function of "I ought to do so-and-so" when it is
used to announce a decision in a case like that of Sartre's pupil.
Its use is plainly to commit oneself, to allow that if I do not do
what I say I ought to do, then I am blameworthy. It is a per-
formatory use of "I ought" in that its use makes one responsible
for performing a particular action where before saying "I ought"
one could not have been held responsible for performing that
action rather than some alternative one. To note this is to bring
out the oddity in Hare's treatment of the "Existentialist's" con-
tribution to his dialogue. For in this nonuniversalizable sense of
"ought" one could never say "You oughtn't" but only "I
oughtn't." To say "You oughtn't" and suppose that you had used
"ought" in this sense would be as odd as to say "You promise"
and suppose that thereby one had committed someone else to a
promise.
Secondly, it might be argued that the very possibility of a
problem such as that of Sartre's pupil presupposes the acceptance
of certain universalizable maxims as moral principles. If Sartre's
pupil had not accepted the maxims "One ought to assist one's
parents when they are in need" and "One ought to assist one's
country when it is in need" there would have been no problem.
What is important is that the clash between two principles need not
be resolved by reformulating one of the principles or formulating a
third one. Certainly this clash could be so resolved. Sartre's pupil
might have acted on the maxim "Duties to one's parents always
have precedence over duties to one's country." Had he done so
he would have legislated not only for his own but for all relevantly
similar situations. But in order to make his own decision he does
not need to so legislate. Now it seems to be a consequence of Hare's
What morality is not 99
position that if the decision between principles is itself to be a
moral decision it must itself rest upon the adoption of a universaliz-
able maxim. This, in the light of Sartre's example, could only be
defended by an a priori restriction on the use of the word "moral."
Such a restriction, however, would not be merely a restriction
upon our use of a word. For to adopt Hare's use of "moral" would
be to permit only one way of settling conflicts of principle (that of
formulating a new principle or reformulating an old one) to be
counted as genuinely a moral solution to a moral problem, while
another way-that of the nonuniversalizable decision ala Sartre-
would be ruled out from the sphere of morality. To do this is
plainly to do more than to offer a descriptive analysis of the
meaning of "moral." It is to draw a line around one area of moral
utterance and behavior and restrict the term to that area.
What one can conclude from this is twofold. First, not all, but
only some, moral valuations are universalizable. What leads Hare
to insist that all are is his exclusive concentration on moral rules.
For rules, whether moral or nonmoral, are normally universal in
scope anyway, just because they are rules. As Mr. Isaiah Berlin has
written in another context, "In so far as rules are general instruc-
tions to act or refrain from acting in certain ways, in specified
circumstances, enjoined upon persons of a specified kind, they
enjoin uniform behavior in identical cases."1 If this is so, then
there is nothing specific to moral valuation in universalizability and
in so far as moral valuations are not expressions of rules they are
not universalizable. Secondly, the exceptions are not simply cases
analogous to that of Sartre's pupil. A whole range of cases can be
envisaged where moral valuations are not universalizable. At the
one extreme would be those instances where in adopting a moral
position someone consciously refrains from legislating for others,
although they might have done so; where a man says, for example,
"I ought to abstain from participation in war, but I cannot
criticize or condemn responsible nonpacifists," but might have
said, "One ought to abstain from participation in war." In such a
case whether to make a universal or a merely personal judgment
is itself a moral problem. The fact that a man might on moral
grounds refuse to legislate for anyone other than himself (perhaps
on the grounds that to do so would be moral arrogance) would by
itself be enough to show that not all moral valuation is univer-
salizable. Or rather that once again this thesis can only be main-
tained by an a priori and quite unjustifiable restriction upon the
word "moral." In other words, a man might conduct his moral
1 "Equality," in Proceedings of the Aristotelian Society (1955-56).
100 Against the Self-Images of the Age
life without the concept of "duty" and substitute for it the concept
of "my duty." But such a private morality would still be a morality.
More commonly, however, nonuniversalizable judgments occur
when a man finds that the concept of "duty ' has limits which
render it useless in certain situations of moral perplexity. Such is
the example of Sartre's pupil. And such are the cases at the other
end of our scale where moral valuations must be nonuniversalizable,
where it is logically impossible to universalize. This is the case with
what the theologians call "works of supererogation." A work of
supererogation is by definition not numbered among the normal
duties of life. Those duties-such things as keeping one's promises
and paying one's debts-are partly characterized by the fact that
the maxims which enjoin them are universalizable. But there are a
great many acts of moral worth which do not come within their
scope: one may be virtuous in the sense in which virtue is
demanded of everyone without being morally heroic. A moral hero,
such as Captain Oates, is one who does more than duty demands.
In the universalizable sense of "ought" it does not therefore make
sense to assert that Captain Oates did what he ought to have done.
To say of a man that he did his duty in performing a work of
supererogation is to contradict oneself. Yet a man may set himself
the task of performing a work of supererogation and commit
himself to it so that he will blame himself if he fails without
finding such a failure in the case of others blameworthy. Such a
man might legitimately say, "I have taken so-and-so as what I
ought to do." And here his valuation cannot, logically cannot, be
universalized.
II
Crucial to the argument so far that universalizability is not a
necessary characteristic of all moral valuation has been the
distinction between first-person and third-person uses of moral
valuation. Before the force of this distinction can be fully under-
stood, however, it is necessary to inquire what the function of
moral valuation may be. The argument of this section will be that
there are a great variety of uses to which moral utterance may be
put, none of which can claim the title of "the" function of moral
valuation. It will be useful to list some of the tasks which even
so familiar a form of moral judgment as "X ought to do Y" may
be set.
1. The expressionof indignationor other violent or mild emotion.
What moralityis not IOI
III
This incomplete catalogue of uses of "ought" in simple sentences
such as "X ought to do Y" has one main point: moral philosophy to
date has been insufficiently lexicographical. Even a partial
enumeration of the differences already noted between first-,
second-, and third-person uses of "ought" (of which that between
a particular first- and a particular third-person use noted in the
discussion of Sartre's example now turns out to be only a parti-
cular case) should make us conscious of the need for a far wider
range of patterns of analysis than any contemporary writer has so
far offered. But, instead of enlarging on this topic here, a possible
reply to the arguments that universalizability is not a necessary
attribute of moral valuation of the form "X ought to do Y", and
that such valuations do not necessarilyhave a prescriptive function,
must be considered. Against these contentions the following
counterargument might be brought.
The essence of moral judgments it might be said is their imper-
sonality. When we judge morally it is at the heart of the matter
that we "do not make exceptions in our own favor" (Kant), that
the moral agent must "depart from his private and particular
situation" (Hume). When the moral agent judges an action he
judges therefore what anyone should do in that or relevantly
similar situations. When he appraises the action of another he
thereby commits himself to saying what anyone and a fortiori
he himself ought to have done. When he decides how to act he
thereby commits himself to an appraisal of any similar action by
anyone else. Thus appraisal, advice, and practical decisions are
inexorably linked together. But of these three, practical decisions
have the primacy; to appraise someone else's action is to say how
he ought to have acted and to give advice is to tell someone else
how to act. Moral language, or at least "ought", is employed par
excellence in guiding action. In this form the argument brings
out the interconnection of the claim that moral judgments
are essentially universalizable and the claim that they are essen-
tially practical and prescriptive. Its force is further brought out
by noting a consequence which Hare has drawn from the con-
clusions of this argument. Hare argues that to say that a man holds
a moral principle is to say that he at least sometimes acts on it. A
man who claims to believe in keeping promises but habitually
breaks them does not in fact hold the principle that one ought to
keep promises, according to Hare. Those who have objected to this
104 Against the Self-Images of the Age
contention have usually pointed to the problem of clxpor:a!or:, have
argued that if Hare were right we would not have the case of the
man whose practice is radically inconsistent with his principles.1 But
this objection takes no cognizance of the way in which the notion
of consistency is built into this argument at the theoretical level.
Take the example of a man who appraises actions by one
standard and guides his own conduct by another. This differs
from the case of the man who is guilty of weakness of will, for such
a man's conduct is consistent with his principles, or rather with
that set of his principles which he uses to guide his conduct.
He merely has two sets of principles. This is sometimes con-
demned by invoking the maxim "Practice what you preach"
which is also, of course, used to condemn weakness of will. We
condemn such a man because and if we disapprove of incon-
sistency between appraisals and principles of conduct. But while
such inconsistency may be morally objectionable, it is not-and
the fact that it can be comprehended to such a degree as to be
found morally objectionable shows that it is not-unintelligible.
Yet in the argument outlined above this is what it must be. For if
the meaning of the appraisal "He ought to have done Y" is even
partly "I ought to have done Y in those circumstances" (inter-
preted as "That is the maxim that would have governed my
conduct" not "That is the maxim by which I would have appraised
my conduct") then the man who asserts that he appraises by one
set of principles but acts by another speaks unintelligibly. In other
words, the view that I am criticizing makes consistency between
appraisals and principles of conduct a logical requirement. That
principles should be so consistent is built into the meaning of
moral words such as "ought." But the demand for consistency
is in fact a moral not a logical requirement. We blame a man for
moral inconsistency perhaps, but we do not find what he says
meaningless. Appraisals and principles of conduct are logically
independent, although in a liberal morality they are required to be
morally interdependent. And now we can understand why univer-
salizability is given such a central place by those philosophers
whose analyses are directed upon the concepts of liberal morality.
For the requirement that everyone shall be judged by the same
standard (the moral counterpart of the political principle that
everyone is to count as one and nobody as more than one) in the
sense that everyone shall judge everyone else by the standard by
which he judges himself is so basic to liberal morality that it is
1 E.g., P. L. Gardiner, "On Assenting to a Moral Principle," Pro-
ceedings of the Aristotelian Society, (1954-55).
What morality is not 105
I
Sometimes in the history of philosophy the defense of a particular
philosophical position and the interpretation of a particular
philosopher become closely identified. This has notoriously
happened more than once in the case of Plato, and lately in
moral philosophy it seems to me to have happened in the case of
Hume. At the center of recent ethical discussion the question of
the relationship between factual assertions and moral judgments has
continually recurred, and the nature of that relationship has
usually been discussed in terms of an unequivocally sharp dis-
tinction between them. In the course of the posing of this question,
the last paragraph of Book III, Part i, Section 1, of Hume's
Treatise has been cited over and over again. This passage is either
quoted in full or at least referred to-and with approval-by
R. M. Hare, 2 A. N. Prior, 3 P. H. Nowell-Smith," and a number
of other writers. Not all contemporary writers, of course, treat
Hume in the same way; a footnote to Stuart Hampshire's paper,
"Some Fallacies in Moral Philosophy," 5 provides an important
exception to the general rule. But very often indeed Hume's
contribution to ethics is treated as if it depended largely on this
one passage, and this passage is accorded an interpretation which
has acquired almost the status of an orthodoxy. Hare has even
spoken of "Hume's Law." 6
1 Reprinted from The Philosophical Review, 1959.
8 The Language of Morals (London and New York: Oxford University
Press, 1952), pp. 29 and 44.
8 Logic and the Basis of Ethics (London and New York: Oxford Univer-
II
To approach the matter obliquely, how can we pass from "is"
to "ought"? In Chapter iv of The Language of Morals, Hare
asserts that a practical conclusion and a fortiori a moral conclusion
is reached syllogistically, the minor premise stating "what we
should in fact be doing if we did one or other of the alternatives
open to us" and the major premise stating a principle of conduct.
This suggests an answer to our question. If you wish to pass
from a factual statement to a moral statement, treat the moral
statement as the conclusion to a syllogism and the factual state-
ment as a minor premise. Then to make the transition all that is
needed is to supply another moral statement as a major premise.
And in a footnote to Chapter iii of Ethics we find Nowell-Smith
doing just this. He quotes the following passage from Bishop
R. C. Mortimer: "The first foundation is the doctrine of God the
Creator. God made us and all the world. Because of that He has an
absolute claim on our obedience. We do not exist in our own right,
but only as His creatures, who ought therefore to do and be what
He desires." 1 On this Nowell-Smith comments: "This argument
requires the premise that a creature ought to obey his creator,
which is itself a moral judgment. So that Christian ethics is not
founded solely on the doctrine that God created us :" 2 That is,
he argues that the inference, "God created us, therefore we ought
to obey him," is defective unless and until it is supplied with a
major premise, "We ought to obey our creator."
I can only make sense of this position by supposing that under-
lying it there is an assumption that arguments must be either
deductive or defective. But this is the very assumption which
underlies Hume's skepticism about induction. And this skepticism
is commonly treated as resting upon, and certainly does rest
upon, a misconceived demand, a demand which P. F. Strawson
has called "the demand that induction shall be shown to be
really a kind of deduction." 3 This is certainly an accurate way of
characterizing Hume's transition from the premise that "there
1 Christian Ethics (London: Black, 1950), p. 7.
2 Op. cit., p. 51.
3 Introduction to Logical Theory (London: Methuen; New York:
Barnes & Noble, 1952), p. 250.
II2 Against the Self-Images of the Age
can be no demonstrative arguments to prove, that those instances
of which we have had no experience resemble those of which we
have had experience" to the conclusion that "it is impossible for
us to satisfy ourselves by our reason, why we should extend that
experience beyond those particular instances which have fallen
under our pbservation." 1 Part of Hume's own point is that to
render inductive arguments deductive is a useless procedure.
We can pass from "The kettle has been on the fire for ten minutes"
to "So it will be boiling by now" (Strawson's example) by way of
writing in some such major premise as "Whenever kettles have
been on the fire for ten minutes, they boil." But if our problem
is that of justifying induction, then this major premise itself
embodies an inductive assertion that stands in need of justification.
For the transition which constitutes the problem has been justified
in the passage from minor premise to conclusion only at the cost
of reappearing, as question-beggingly as ever, within the major
premise. To fall back on some yet more general assertion as a
premise from which "Whenever kettles have been on the fire for
ten minutes they boil" could be derived would merely remove the
problem one stage farther and would be to embark on a regress,
possibly infinite and certainly pointless.
If then it is pointless to present inductive arguments as deduc-
tive, what special reason is there in the case of moral arguments for
attempting to present them as deductive? If men arguing about
morality, as Bishop Mortimer is arguing, pass from "God made
us" to "We ought to obey God," why should we assume that the
transition must be an entailment? I suspect that our inclination
to do this may be that we fear the alternative. Hare suggests that
the alternative to his view is "that although, in the strict sense of the
word, I have indeed shown that moral judgments and imperatives
cannot be entailed by factual premises, yet there is some looser
relation than entailment which holds between them." I agree with
Hare in finding the doctrine of what he calls "loose" forms of
inference objectionable; although I cannot indeed find this
doctrine present in, for example, Professor S. E. Toulmin's The
Place of Reason in Ethics which Hare purports to be criticizing.
And certainly entailment relations must have a place in moral
argument, as they do in scientific argument. But since there are
important steps in scientific argument which are not entailments,
it might be thought that to insist that the relation between factual
statements and moral conclusions be deductive or nonexistent
1 Treatise, I, iii, 6; (Selby-Bigge, pp. 89 and 91).
Hume on "is" and "ought" 113
would be likely to hinder us in elucidating the character of moral
arguments.
How does this bear on the interpretation of Hume? It might
be held that, since Hume holds in some passages on induction
at least that arguments are deductive or defective, we could
reasonably expect him to maintain that since factual premises
cannot entail moral conclusions-as they certainly cannot-there
can be no connections between factual statements and moral
judgments (other perhaps than psychological connections). But
at this point all I am suggesting is that our contemporary dis-
approval of Hume on induction makes our contemporary approval
of what we take to be Hume on facts and norms seem odd. It is
only now that I want to ask whether-just as Hume's attitude to
induction is much more complex than appears in his more scep-
tical moments and is therefore liable to misinterpretation-his
remarks on "is" and "ought" are not only liable to receive but have
actually received a wrong interpretation.
III
The approach will still be oblique. What I want to suggest next
is that if Hume does affirm the impossibility of deriving an
"ought" from an "is" then he is the first to perform this particular
impossibility. But before I proceed to do this, one general remark
is worth making. It would be very odd if Hume did affirm the
logical irrelevance of facts to moral judgments, for the whole
difference in atmosphere-and it is very marked-between his
discussion of morality and those of, for example, Hare and
Nowell-Smith springs from his interest in the facts of morality.
His work is full of anthropological and sociological remarks,
remarks sometimes ascribed by commentators to the confusion
between logic and psychology with which Hume is so often
credited. Whether Hume is in general guilty of this confusion is
outside the scope of this essay to discuss. But so far as his moral
theory is concerned, the sociological comments have a necessary
place in the whole structure of argument.
Consider, for example, Hume's account of justice. To call an
act "just" or "unjust" is to say that it falls under a rule. A single
act of justice may well be contrary to either private or public
interest or both.
IV
What I have so far argued is that Hume himself derives "ought"
1 Op. cit., Selby-Bigge, p. 498.
Hume on "is" and "ought" 117
from "is" in his account of justice. Is he then inconsistent with
his own doctrine in that famous passage? Someone might try to
save Hume's consistency by pointing out that the derivation of
"ought" from "is" in the section on justice is not an entailment
and that all Hume is denying is that "is" statements can entail
"ought" statements, and that this is quite correct. But to say this
would be to misunderstand the passage. For I now want to argue
that in fact Hume's positive suggestions on moral theory are
actually an answer to a question posed in the "is" and "ought"
passage, and that that passage has nothing to do with the point
about entailment at all. The arguments here are twofold.
First, Hume does not actually say that one cannot pass from an
"is" to an "ought" but only that it "seems altogether incon-
ceivable" how this can be done. We have all been brought up
to believe in Hume's irony so thoroughly that it may occasionally
be necessary to remind ourselves that Hume need not necessarily
mean more or other than he says. Indeed the rhetorical and
slightly ironical tone of the passage renders it all the more am-
biguous. When Hume asks how what seems altogether incon-
ceivable may be brought about, he may be taken to be suggesting
either that it simply cannot be brought about or that it cannot
be brought about in the way in which "every system of morality
which I have hitherto met with" has brought it about. In any case
it would be odd if Hume thought that "observations concerning
human affairs" necessarily could not lead on to moral judgments
since such observations are constantly so used by Hume himself.
Secondly, the force of the passage as it is commonly taken
depends on what seems to be its manifest truth: "is" cannot entail
"ought." But the notion of entailment is read into the passage. The
word Hume uses is "deduction." We might well use this word as a
synonym for entailment, and even as early as Richard Price's
moral writings it is certainly so used. But is it used thus by Hume?
The first interesting feature of Hume's use of the word is its
extreme rarity in his writings. When he speaks of what we should
call "deductive arguments" he always uses the term "demon-
strative arguments." The word "deduction" and its cognates have
no entry in Selby-Bigge's indexes at all, so that its isolated occur-
rence in this passage at least stands in need of interpretation.
The entries under "deduction" and "deduce" in the Oxford
English Dic_tionarymake it quite clear that in ordinary eighteenth-
century use these were likely to be synonyms rather for "inference"
and "infer" than for "entailment" and "entail." Was this Hume's
usage? In the essay entitled "That Politics may be Reduced to a
n8 Against the Self-Images ofthe Age
Science," Hume writes, "So great is the force of laws, and of
particular forms of government, and so little dependence have
they on the humors and tempers of men, that consequences
almost as general and certain may sometimes be deduced from
them as any which the mathematical sciences afford us." 1 Clearly,
to read "be entailed by" for "deduced from" in this passage
would be very odd. The reference to mathematics might indeed
mislead us momentarily into supposing Hume to be speaking of
"entailment." But the very first example in which Hume draws a
deduction makes it clear how he is using the term. From the
example of the Roman republic which gave the whole legislative
power to the people without allowing a negative voice either to
the nobility or the consuls and so ended up in anarchy, Hume
concludes in general terms that "Such are the effects of democracy
without a representative." That is, Hume uses past political
instances to support political generalizations in an ordinary
inductive argument, and he uses the term "deduce" in speaking
of this type of argument. "Deduction" therefore must mean
"inference" and cannot mean "entailment."
Hume, then, in the celebrated passage does not mention en-
tailment. What he does is to ask how and if moral rules may be
inferred from factual statements, and in the rest of Book III of
the Treatise he provides an answer to his own question.
VI
The argument of this essay is incomplete in three different ways.
First, it is of certain interest to relate Hume's argument to con-
temporary controversies. On this I will note only as a matter of
academic interest that there is at least one recent argument in
which Hume has been recruited on the wrong side. In the dis-
cussion on moral argument between Hare and Toulmin, 2 Hare has
invoked the name of Hume on the side of his contention that
factual statements can appear in moral arguments only as minor
premises under the aegis of major premises which are statements of
moral principle and against Toulmin's contention that moral
arguments are nondeductive. But if I have reread Hume on "is"
1 Preface.
1 The Language of Morals, p. 45; Philosophical Quaf'terly, I (1950-51),
p. 372; and Philosophy, XXXI (1956), p. 65.
Hume on "is" and "ought" 123
and "ought" correctly, then the difference between what Hume has
been thought to assert and what Hume really asserted is very much
the difference between Hare and Toulmin. And Hume is in fact as
decisively on Toulmin's side as he has been supposed to be on
Hare's.
Secondly, the proper elucidation of this passage would require
that its interpretation be linked to an interpretation of Hume's
moral philosophy as a whole. Here I will only say that such a thesis
of Hume's as that if all factual disagreement were resolved, no
moral disagreements would remain, falls into place in the general
structure of Hume's ethics if this interpretation of the "is" and
"ought" passage is accepted; but on the standard interpretation it
remains an odd and inexplicable belief of Hume's. But to pursue
this and a large variety of related topics would be to pass beyond
the scope of this essay.
Finally, however, I want to suggest that part of the importance
of the interpretation of Hume which I have offered in this essay
lies in the way that it enables us to place Hume's ethics in general
and the "is" and "ought" passage in particular in the far wider
context of the history of ethics. For I think that Hume stands at a
turning point in that history and that the accepted interpretation of
the "is" and "ought" passage has obscured his role. What I mean
by this I can indicate only in a highly schematic and speculative
way. Any attempt to write the history of ethics in a paragraph is
bound to have a I066 and All That quality about it. But even if the
paragraph that follows is a caricature it may assist in an under-
standing of that which it caricatures.
One way of seeing the history of ethics is this. The Greek moral
tradition asserted-no doubt with many reservations at times--an
essential connection between "good" and "good for," between
virtue and desire. One cannot, for Aristotle, do ethics without
doing moral psychology; one cannot understand what a virtue is
without understanding it as something a man could possess and as
something related to human happiness. Morality, to be intelligible,
must be understood as grounded in human nature. The Middle
Ages preserves this way of looking at ethics. Certainly there is a
new element of divine commandment to be reckoned with. But the
God who commands you also created you and His commandments
are such as it befits your nature to obey. So an Aristotelian moral
psychology and a Christian view of the moral law are synthesized
even if somewhat unsatisfactorily in Thomist ethics. But the
Protestant Reformation changes this. First, because human beings
are totally corrupt their nature cannot be a foundation for true
124 Against the Self-Images of the Age
morality. And next because men cannot judge God, we obey God's
commandments not because God is good but simply because He is
God. So the moral law is a collection of arbitrary fiats unconnected
with anything we may want or desire. Miss G. E. M. Anscombe has
recently suggested that the notion of a morality of law was
effectively dropped by the Reformers ;1 I should have thought that
there were good grounds for asserting that a morality of law-and-
nothing-else was introduced by them. Against the Protestants
Hume reasserted the founding of morality on human nature. The
attempt to make Hume a defender of the autonomy of ethics is
likely to conceal his difference from Kant, whose moral philosophy
is, from one point of view, the natural outcome of the Protestant
position. And the virtue of Hume's ethics, like that of Aristotle
and unlike that of Kant, is that it seeks to preserve morality as
something psychologically intelligible. For the tradition which
upholds the autonomy of ethics from Kant to Moore to Hare,
moral principles are somehow self-explicable; they are logically
independent of any assertions about human nature. Hume has been
too often presented recently as an adherent of this tradition.
Whether we see him as such or whether we see him as the last
representative of another and older tradition hinges largely on how
we take what he says about "is" and "ought."
I
Jespersen says that "The imperative is used in requests, which
according to circumstances may range from brusque commands to
humble entreaties, the tone generally serving as a key to the exact
meaning." 2 Hare says that "An indicative sentence is used for
1 Reprinted from The Journal of Philosophy, 1965.
z Otto Jespersen, Essentials of English Grammar (London: Allen &
Unwin; New York: Holt, Rinehart and Winston, 1933), p. 294. In A
Modern English Grammar on Historical Principles (Heidelberg: C. Winter,
7 vols., 1909-49; London: Allen & Unwin, 1927-50) Jespersen adds
wishes and permissions to requests.
125
126 Against the Self-Images of the Age
telling someone that something is the case; an imperative is not- it
is used for telling someone to make something the case." 1 This
concurrence of grammarian and philosopher, however, still leaves
one in some doubt. Both seem to assert that the class of imperative
sentences may be segregated by identifying their use or function.
But when Fanny Farmer says, "Take six eggs," she certainly
neither commands nor entreats and does not even request; and
when in giving an order an officer says, "The following will report
at the Guard Room at 18.00 hours: Smith, Jones, Robinson," an
indicative is being used to tell someone to do something. These
elementary examples throw an important initial doubt on whether
Jespersen and Hare have characterized the relationship between
grammatical form and linguistic function correctly. In order to
strengthen this doubt I want to inquire how, in trying to under-
stand the unfamiliar language of an alien culture, we might dis-
cover whether that language contained imperative sentences and
how to identify them.
The first point to be noted-from the earlier examples-is that
imperative sentences may be used to serve more than one purpose.
They are of course commonly used to tell some specific person to do
something; but they are also commonly used to tell anyone how to
do something. That the same form should serve both purposes is
easily intelligible; for a set of instructions in a cookbook or a
carpentry manual, when recited by a mistress to her servant or by
a master carpenter to an apprentice on a specific occasion as
directions for the performance of a specific task, becomes a set of
orders or requests. It remains true that the existence of the single
grammatical form in virtue of which this is possible is a contingent
feature of English, Latin, and other languages. So the first pre-
caution to be taken in examining an unfamiliar language would be
to inquire whether the two functions of telling someone to do some-
thing and of telling how to do something are served by a single form.
A second preliminary point is that, if attention were concentrated
on telling to, we should have to be careful to distinguish between
imperative sentences and sentences not of imperative form which
are expressions of desires or wishes, translatable as "I want you to
do ... " or "Would that you would do .... " Both classes of
sentence would be employed in utterances in which the naming of
an action to some specified person or range of persons would often
be followed by the performance of the action by those to whom
utterance was addressed, and the action would be performed
1 R. M. Hare, The Language of Morals (London and New York:
Oxford University Press, 1952), p. 5.
Imperatives, reasonsfor action, and morals 127
precisely because of what had just been said. But the reason for the
performance of the action would be quite different in the one case
from the other. "I did it because you wanted me to" is quite dif-
ferent from "I did it because you told me to." None the less, those
imperatives which are used to express requests are often inter-
changeable with expressions of a want. "I would like a lemonade,
please" can be interchanged with "Get me a lemonade, please,"
and the latter can thus function as an expression of a want.
This again is easily intelligible, for "The primitive sign of
wanting is trying to get," 1 and the utterance of an imperative as
either command or request is one form of trying to get. But it
follows that the segregation of genuinely imperative sentences
from sentences of the form "I want ... " is all the more difficult.
Nonetheless, this segregation is essential, both for the reason
already given-that the imperative usually affords the person to
whom it is uttered a different kind of reason for action from that
afforded by the expression of a desire-and also because the
imperative, when it is used to tell someone to do something, need
not in any way express the particular desire or wish of the speaker.
The bored officeron parade may in fact hope that when he issues the
command "Present arms!" all the troops will fall off their horses.
It follows from all this that the isolation of an imperatival form
in a language would be the isolation of a form available for a given
and quite wide range of functions, but that these functions do not
have any obvious unity, although each is related intelligibly to at
least one of the others. From this it seems to follow both that we
can have no a priori reason for expecting the occurrence of such a
grammatical form in any other language and also that in our own
language we cannot use the imperatival form to pick out a homo-
geneous and well-defined class of utterances. It will be at the very
least highly misleading to begin with the concept of an imperative
and then to explain the nature of, for example, moral injunctions
or some other use of language by comparing them to imperatives.
For all that is clearly segregated by the concept of an imperative is
a certain grammatical form. To emphasize this point, it is perhaps
worth while to examine a little further the relations between
telling to, telling how, and telling that.
The introduction of telling how would itself be sufficient to
destroy any neat pairing off of imperatives with telling to and
indicatives with telling that. The instructions by means of which
someone is told how to bake a cake or make a cupboard may equally
1 G. E. M. Anscombe, Intention (Oxford: Basil Blackwell; Ithaca,
II
Philosophical inquiries into the relation of imperatives to reasoning
have tended to concentrate on the deductive relationships into
which imperative sentences may enter and have been conducted
with an eye to the alleged large dichotomies of fact and value,
describing and prescribing, and so on. Less ambitiously, I want to
make some more elementary points about the relationship between
imperatival injunctions and the reasons for action from which they
derive their specific character as threats, warnings, and the like on
the one hand and orders and commands on the other. The first is
that someone to whom an injunction in the form of a warning or a
threat is addressed may take up three attitudes toward it: he may
accept it, either by doing as he is enjoined to do or by taking pre-
cautions against the consequences of not so doing; he may refuse
to believe the statement that asserts the possibility or probability
of such consequences and, the.refore, not act as he is enjoined to do;
or again, he may believe the statement and yet think that it does not
adequately warrant obedience to the injunction. It is upon the
first and third cases that I wish to concentrate attention in order to
ask what may be the difference between a warranted and an
unwarranted injunction. Suppose a man threatened with dire
consequences if he does not take one course of action is advised
by a friend to take another incompatible course of action. The
blackmailer says, "Forge the check," saying or implying "or else
I will ruin your reputation." The friend says, "Don't forge the
check. You will always regret it." (Note incidentally that in this
example the imperatives are being used in an attempt to persuade.
Imperatives need not so be used. A man who seeks to advise need
Imperatives, reasonsfor action, and morals 131
III
We are now in a position to summarize some of the difficulties that
stand in the way of the protagonists of any theory purporting to
assimilate moral utterances to imperatives. First of all, since
it was made clear that the imperatival form does not of itself
segregate any well-defined class of utterances, it is clear that some
better defined class of utterances that can be expressed by means of
imperatives would have to be picked out to provide a comparison
with moral utterances. Secondly, it was also made clear that, where
the class of utterances that can be characterized as instances of
telling to rather than of telling that or of tellinghow are concerned,
we must ask for some further characterization beyond that of
telling to. Now what has been said about commands and orders
makes it highly implausible that moral utterances could be
illuminated by comparison with them; it is, therefore, all the more
plausible that, if moral utterances are a form of or resemble
imperatival injunctions, the class of imperatival injunctions to be
examined is that which includes threats, warnings, and pieces of
134 Against the Self-Imagesof the Age
advice. But the account of this class of injunctions which has
already been given would appear to make any assimilation of
moral utterances to imperatives highly implausible, since that
account at least suggeststhat in moral discourse perfectly ordinary
and genuine imperatives may play a part and that the distinguishing
moral components of moral utterance will be found in the statement
of the reasons backing up the imperatival injunction and not in the
imperative itself. For just this was found to be the case with the two
types of "ought" statements so far briefly examined.
To this it may be replied that neither of these "oughts" is the
moral "ought." The first might be ruled out on the ground that
it is the "ought" of hypothetical and prudential injunctions
rather than that of moral injunctions, the second on the grounds
that it is the "ought" of legally and socially established codes, not
the "ought" of individual moral principle. The moral "ought"
is, therefore, to be sought elsewhere, and, when it is found, it will
turn out that "You ought to do X and not Y, but do Y and not
X"-is not a possible piece of advice if the "ought" is the moral
"ought." For the moral "ought" will have an imperatival character,
such that anyone who assents to a moral injunction assents to the
corresponding imperative. Indeed it has been held that to assent
sincerely to a moral injunction entails not only assenting to the
corresponding imperative but also acting on it when occasion
requires; and yet clearly, in the view that I have outlined, I may
sincerely assent to and believe the statement that you ought to do
something but yet tell you by means of an imperatival injunction
to do something else, and I may even sincerely assent to and
believe the statement that I ought to do something but find reasons
for doing something else.
It remains, therefore, to hunt out the moral "ought" and see if it
has a different relation to the imperative mood. Now it might
appear that this was bound to be a hopeless quest. For the first
type of "ought" was in fact that characteristic of those teleological
moralities in which the criterion of what a man ought to do is what
will enable him to fare well or badly, and some concept of what it
is for a man to fare well or badly underpins the morality. So the
heirs of Aristotle. And the second type of "ought" is characteristic
of these moralities in which belief in a divine law is fundamental,
and moral rules have a status similar to the status of legal statutes.
So the heirs of the Stoics and the Torah. Since these "oughts"
are therefore at home in moralities of such importance, how could
some other "ought" claim the title of "the" moral "ought" ? But
to pursue this line of argument might be to miss the point.
Imperatives, reasonsfor action, and morals 135
Suppose a society in which there was a moral ethos of either the
Aristotelian or the Judaeo-Christian kind or perhaps some
synthesis of the two. The uses of "ought" in such a society would
be as I have outlined, and its relation to imperatival injunctions
would also be as I have outlined. But now suppose also that belief
in the divine law declines and that the formerly shared conception
of what human well-being consisted in is no longer shared.
None the less, people continue to use the word "ought," but in new
contexts. Sometimes, indeed, they ask skeptically whether they
ought to obey the divine law or whether one ought to preserve
former conceptions of human well-being. The "ought" of these
inquiries clearly cannot be either of the "oughts" previously
current. When "ought" is now used to reinforce an imperatival
injunction, it will add little but emotive force. When used by
itself: "You ought to do X and not Y," what force is there left
for it to have but an imperatival force?
In such a situation emotivist and prescriptivist theories, indeed
any theory of a kind which assimilated "ought" to imperatival
injunctions, would tend to flourish. Nor would these theories be
wholly incorrect; for they would be describing correctly the role
in language which a certain class of sentences had assumed when
cut off from the background of beliefs necessary for them to be
understood as they had been in the past. Moreover, such theories
might well, if they enjoyed a vogue, help to propagate the linguistic
usages of which they furnished an analysis. But we should expect,
with such a degree of innovation, considerable linguistic strain.
The framework of our language would, one would expect, be such
that the claim that the form, "You ought to do X and not Y, but
do Y and not X," does not make sense or indicates insincerity or
embodies a contradiction would be implausible in all sorts of ways.
None the less, the inclination to make this claim, in order to
explain correctly how "ought" was now often being used, would
be widespread.
Just this may well be our situation. It is at least a plausible
hypothesis. If it is correct, then simply to conclude that the case for
assimilating moral utterances to imperatives could not be made
out would be to ignore changes that social history has brought
about in the linguistic relationships. And this would reinforce the
suggestion, which has been apparent throughout this essay, that
to attempt to characterize a given class of utterances by means of
an appeal to grammatical form is one of the less happy ways of
making use of linguistic classifications in the course of philosophical
inquiry.
"Ought"
I
Not only the suggestion that different societies have had widely
different moral beliefs, but also the more radical suggestion that
the conceptual schemes embodied in their moralities have differed
widely, would appear as a banal truism to any anthropologist.
From Vico to Karl Marx, moral philosophers too appeared willing
to entertain this prosaic suggestion. But the notion of a single,
unvarying conceptual structure for morality dies hard; and from
the eighteenth century to this day, the English utilitarians and
idealists, logical empiricists and analytical philosophers, have all
been willing to discuss moral philosophy on the assumption that
there was something to be called "the moral consciousness" or,
in a later idiom, "the language of morals." The questions "Whose
moral consciousness?" or "Which language?" have rarely, if ever,
been raised. Both a cause and a consequence of this situation is the
extremely short method with examples adopted by many writers.
Since the question "But is this how anyone in particular really
talks, and if so who?" is not considered relevant, since we are
ostensibly concerned with what can or what cannot be said and not
with what is as a matter of fact said, a self-indulgent imaginative
liberty is sometimes allowed to reign. Moral philosophers are
often able to discern and to decry this liberty in their immediate
predecessors : it is not unfashionable to pillory the abstract and
misleading character of Ross's use of perplexity over a promise to
return a book which I have borrowed as a paradigmatic example
of a moral dilemma or of Moore's famous choice between two
uninhabited worlds. But if the abbreviation of their own examples
by moral philosophers is a liberty, the short shrift which they
administer to the examples of others is more a form of intellectual
violence. I want to try and avoid such extreme methods in a
consideration of two recent attempts to characterize what have
136
"Ought" 1 37
II
By prescriptivism I understand the following set of positions:
1. That the utterance of a moral judgment commits the utterer
to the utterance of a corresponding imperatival injunction in
such a way that "You ought to do this, but don't" or "I ought to
do this, but don't" (addressed to me) are inconsistent expressions.
In cases where "You ought to do this, but don't" is not to be
construed as inconsistent, then the "ought" is some secondary,
derivative, perhaps inverted-commas use of "ought," perhaps with
the meaning "People in general think you ought, but don't."
2. That no moral, indeed no prescriptive, use of "ought" can
appear in a categorical judgment that is the conclusion of a valid
argument, if the premises of that argument are all factual assertions.
For such an argument to be valid, at least one of the premises must
contain a prescriptive "ought." No "ought" from "is."
3. That the only limitations upon our choice of moral
principles are those of syntactical form and of consistency; we
cannot choose a moral principle which we should not be prepared
to apply equally to others and to ourselves in relevantly similar
circumstances; and that in moral justification our chain of
reasoning will bring us at last to a set of first principles ; these
cannot be supported further by reasons, but must be adhered to
by choice.
4. That to call something good may often be to apply con-
ventionally accepted criteria for the goodness of whatever it is that
is called good. But it is also-except in inverted-commas cases-to
138 Against the Self-Images of the Age
endorse the criteria, if they are conventional, and this endorsement
expresses the agent's own choice of criteria.
5. That in ordinary language there are many expressions in
which, either in certain contexts or always, a prescriptive or
evaluative and a descriptive component are welded together. The
use of these creates an illusion that the applicability of a particular
description entails the making of a particular evaluative judgment.
But, in fact, it is always possible to separate the two components ;
their relationship is purely contingent, external, and conventional.
By naturalism, I understand the followil).gset of positions :
I. That evaluative conclusions may be validly derived from
factual premises by virtue of the criteria for the correct application
of certain evaluative expressions; that such criteria are factual and
that the relevant class of expressions includes the virtue words.
From the fact that you acted in such and such a way it follows
that you are courageous, and in being courageous you behaved as
a man ought to behave.
2. That the chain of justifications for holding that someone
ought to do something will lead back, not to an ultimate assertion
of principle resting upon a choice which can have no further
justification, but to an assertion that behavior of the kind in
question is a realization of or is productive of some recognized
human good, such as health and pleasure.
3. That the class of such goods can be delimited; that the
criteria for something's being a good are independent of the agent's
choices, and that therefore what I ought to do does not depend
on my choices, although what I am going to do may so depend.
4. That the question of what I ought (in the fullest moral sense)
to do can be raised and settled by me without the question of what
I am going to do being so raised, let alone settled.
5. That factual or descriptive and evaluative or prescriptive
are not mutually exclusive predicates of judgments, and that
a judgment which is both factual and evaluative is not a judgment
which combines two separable components; to say that it is
evaluative is to speak about the point of making it, while to say
that it is factual is to speak about how its truth or falsity is to be
determined.
The strategies and counter-strategies used by each of these
parties against the other are parallel ones, and their ritual pro-
cedures might be set out as follows. First, a counter-example is
cited by one party; secondly, the other produces its own account of
the counter-example; thirdly, the first party points out that some
crucial distinction is omitted by this account; fourthly, the other
"Ought" 139
party either accepts this omission with all its consequences, or
gives its own quite different account of the distinction. Now an
unlimited willingness to accept the consequences of their own
point of view, which characterizes the leading protagonists on both
sides, leads to an increasing doubt on the part of an observer as to
what the subject matter of this controversy in fact is. I propose
to follow through the course of two of these ritual engagements to
illustrate this point, one starting-point being taken from each side
of the controversy. But it may help to exhibit the inaccessibility of
each party to the other's polemics, if I follow through two argu-
ments in which essentially the same issue is at stake. We can put
this issue as follows. The individual agent comes upon the moral
scene with a set of wants and interests. He finds the members of
his society speaking an evaluative language such that the normal
usage is expressive of certain established standards. Naturalist
and prescriptivist are both prepared to agree that every established
society will speak a moral and evaluative language which does
express the conventionally accepted standards. What they dis-
agree about is whether this is purely a contingent union of dis-
parate elements, so that the key terms could be purged of their
descriptive meaning and so of their particular adherence to a
particular way of life, or whether it is impossible at the most
fundamental level to perform this task. Can the individual agent
construct a set of moral standards that are genuinely his own, or
can he only appeal to something that may be called a standard,
if he agrees to speak the language that he finds? In another idiom,
do we make or do we discover our values?
The central naturalist charge against prescriptivism in this
area of the controversy is that the prescriptivist cannot in impor-
tant cases distinguish evaluative judgments from the expression of
private wants and preferences. The central prescriptivist charge
against naturalism is that the naturalist cannot give any account
of evaluative innovation, that yet the possibility of such innovation
is always present. The naturalist, to spell it out more fully,
asserts that we cannot but consider certain dispositions virtues.
To this the prescriptivist may reply that a situation in which the
meanings of the moral expressions in our language guarantee that
the standards by which we decide what is or is not a virtue are
unalterable is a I984 situation; and moral language does not have
to be like Newspeak nor is it commonly likeNewspeak. The natural-
ist reply in turn will be that any particular moral scheme may be
criticized or transcended by reference to universal human goods,
such as happiness. To this the prescriptivist will rejoin that if
Against the Self-Images of the Age
"happiness" is invoked in this way it is as a prescriptive and not
as a descriptive term, and that the agent has a choice of whether
to invoke happiness in this way or not. And so on.
Or to begin the argument from the other side, the prescriptivist
asserts that for our ultimate principles we can give no justificatory
reasons which are other than a specification of what it would be
to live by such principles. To this the naturalist replies that if this
is so, the distinction between the assertion of a moral principle
and the expression of a private want or preference is obliterated.
The prescriptivist will deny this on the grounds that the expres-
sion of a want or preference is not universalizable; even if the agent
does as a matter of contingent fact want both himself to do
such-and-such and everyone else in relevantly similar circum-
stances to do such-and-such, this is not the same as being logically
committed, by one's adherence to the view that one ought oneself
to do such-and-such, to the view that so ought everyone else in
relevantly similar circumstances. The naturalist may admit this
distinction but will insist that there is another distinction between
the assertion of a moral principle and the expression of a want or
preference which the prescriptivist has obliterated. The prescrip-
tivist will reply that this further distinction-that between reason-
supported assertions and non-reason-supported assertions perhaps
-can be made in every case except that of fundamental moral
principles. And so on. Ad na:useam.
I do not doubt that both prescriptivists and naturalists will be
extremely dissatisfied with this portrayal; and there has been no
lack of peacemakers attempting to patch up things between them.
But I would like to suggest that no conclusive argument is found
at any point in these exchanges; conclusive, that is, in terms other
than those of the party that propounds them. And this raises the
interesting question: what are they arguing about? What is the
subject matter of the dispute? Ostensibly the language of morals,
or moral concepts. But moral utterance and moral practice might
be thought to pre-exist philosophical theorizing about morality,
to provide an independent subject matter, so that philosophical
theories would be tested by comparing them with the facts of
moral utterance and practice. To which the reply may be twofold:
first, that some past philosophical theorizing has played at least
some part in shaping later moral concepts, so that moral utterance
is not entirely innocent of philosophical preconceptions; and
secondly that in order to pick out moral utterance and practice (as
contrasted with legal or scientific, say) or in order to pick out
evaluative utterance and practice (as contrasted with predictive,
"Ought" 141
say), one already has to have a criterion and such a criterion will
imply more or less of a theory. The first point one may concede at
once. The force of the second may be brought out by considering
a parallel case.
Clearly, in picking out the legal rules of a particular community
from its other rules we need to approach the social facts with a
criterion; and this criterion will be more or less theory-laden.
When legal theorists, for example, began to examine law in certain
primitive societies, some of them concluded from the fact that
certain factors hitherto considered essential factors of any legal
system whatsoever were absent from the institutional arrange-
ments of these societies that such societies possessed no law at all.
This conclusion, however, totally obscured the genuine con-
tinuity between the sanction-backed public rules of these societies
and the later legal codes which had sometimes grown out of them.
This fact provided grounds for concluding not that such primitive
societies had no legal system, but that the definition of law and
the corresponding theories of law needed to be amended to
accommodate the new facts. This example shows plainly that we
are not necessarily the prisoners of our criteria or theories, even
in cases where we cannot segregate the relevant class of facts
without some initial minimal theoretical commitment.
It follows that if the argument between prescriptivism and
naturalism is not to be an empty and pointless contest, which has
by the very virtuosity of the contestants in the performance of
the task of redescription been deprived of that independent
subject matter, the characterization of which was the sole point
of the whole enterprise, one prerequisite is that as far as possible
both theories are matched against the facts, so far as these can be
independently delineated, and the tendency to redescribe the
facts in accordance with the requirements of the rival theories
must be curbed as far as possible. It is not too difficult to see how
this might be done. It is by approaching the linguistic facts at
first as much as possible in the mode of the lexicographer rather
than of the philosopher; by next setting the linguistic facts in
their social contexts; and finally by asking whether this does not
enable us to discriminate, in relation to the theories of both pre-
scriptivism and naturalism, the types of moral situation of which
each doctrine is the natural and convincing explanation and
analysis from the types of moral situation which one or the other
doctrine has to distort. In so doing we shall treat these doctrines
as hypotheses, which invoke a stylized model of argument to
explain the actual patterns of moral speech and controversy.
Against the Self-Images of the Age
The examples which I shall take are all stages in the history of
the word "ought" and of words in other languages translatable
into English as "ought".
III
"Ought'' and "should" are sometimes used interchangeably in
contemporary English. But "should" often has a subjunctive role
which "ought" never has. ("If you should not hear from me .... ")
Both "ought" and "should" can be used to give advice on any
topic whatsoever; both are sometimes used in such a way that
moral injunctions or advice are distinguished from either prudence
or expediency. "You ought to do so-and-so; how you would fare
if you did is another matter." This use of "ought" or "should"
is the latest in time to emerge. What precedes it is the general
advice-giving "ought" of hypothetical imperatives, whether of
skill or of prudence. If we work backward in time we shall find
that "the moral 'ought' " which is used to express the claims of
Duty (in the singular and with a capital letter), although it appears
prominently in the utterances of Kant, the Duke of Wellington,
George Eliot, and Mr. Gladstone (who wrote of "The two great
ideas of the divine will, and of the Ought, or duty ... "), does not
seem to appear before the eighteenth century.
If we are to properly distinguish the categorical "ought" of
final moral appeal from the "ought" of hypothetically expressed
general advice-giving, it must be done by citing more than the
logical distinction between categorical and hypothetical judgments.
For the general advice-giving "ought" may of course be used in
categorical expressions of advice: not, "You ought if ... " but
"You ought, since .... " What distinguishes the "ought" of final
moral appeal is that it can be used without reason-giving of any
kind, that its force does not depend upon the force of some
attached condition, whether fulfilled or unfulfilled, neither "if"
nor "since," but "Why ought you?" ... "You just ought."
The general advice-giving "ought" is in use from the high
Middle Ages onward. It has to be distinguished from the use in
which "ought" is equated with, indeed is the same word as,
"owe." This "ought" is used both of owing money, and also of
owing services by virtue of occupational status or kinship. "A man
ought to be ... " can mean "A man is owed the position of
being .... " (So Wyntoun writes that "Robert the Brus, Erle of
Karagh, aucht to succeed to be kynrike.") The earliest recorded
"Ought" 143
use of "ought" ("ahte") in English translates "debebat" (in the
sense of "owe" in the Vulgate) by "ahte to zeldanne." We thus
discover three stages in the use of "ought": a first in which "ought"
and "owe" are indistinguishable; a second in which "ought" has
become an auxiliary verb, useable with an infinitive to give advice;
and a third in which the use of "ought" has become unconditional.
We can now place expressions from other languages commonly
translated into English as "ought" in terms of these three stages.
The Greek 8e:i:vand its cognates can be used to express both the
first and the second stage, both what it is incumbent upon a man
to do as a such-and-such, and what a man ought to do if he wants
such-and-such; but not the third stage. The Icelandic "skyldr"
(ancestor of "should") never seems to go beyond the first stage.
It would seem therefore most illuminating to look at the Icelandic
and Greek social contexts for these linguistic facts, in order to
observe first the use of an "ought" that never goes beyond the
first stage and then the use of an "ought" that never goes beyond
the second.
IV
In the society of the Norse sagas, the rules defining social roles and
the obligations attaching to them include the rules of vendetta.
Obligation is tied to kinship so closely that a near kinsman is a
"skyldr fraendi"; and "o-skyldr," which might be naturally, if
clumsily, translated as "not connected to one by way of obligation,"
means "unrelated." In the saga of Gisli the Soursop, Vestein is
killed either by Thorkell or by Thorkell's emissary, and Thorkell
is killed in turn by Vestein's sons. Bork then says, "More I should
take up Thorkell's case than anyone else since he was my brother-
in-law." It follows from the facts that Thorkell has been killed
and that Bork is Thorkell's close kinsman, that Bork ought to
avenge Thorkell's death. What gives the rules by virtue of which
Bork cannot avoid this conclusion their authority is simply the
recognition of these rules by the entire community. Which rules
the community so recognizes and what these rules prescribe are
questions of fact. If we pursue the chain of reasoning by which a
moral conclusion is justified in such a society, it will run as fol-
lows: You ought to kill so-and-so. Why? Because he killed so-and-
so, who is your nearest kinsman. Why because he killed my nearest
kinsman, ought I to kill him? The rules so prescribe. Here the
chain of justifications terminates. There is no way for me to ask
144 Against the Self-Images of the Age
why I ought to obey the rules. Notice that each "Why?" is
answered by a factual assertion.
It would be equally distorting to try to give either a prescrip-
tivist or a naturalistic account of this morality. Suppose, for
instance, that a prescriptivist were to argue that as a matter of
contingent fact agents in such a society may infer the conclusion
"I ought to do such-and-such" from the premise "The rules
prescribe such-and-such" and so appear to derive an "ought"
from an "is"; but that they can do so only by virtue of tacitly
presupposing some unstated major premise, such as: "We ought
always to do what the rules prescribe." The historical facts about
Icelandic society have to be interpreted in the light of our a
priori understanding of the forms of valid arguments. But the
prescriptivist who tries to argue in this way is forced to misrepre-
sent the conceptual scheme of Icelandic morality. For on his own
terms, if the transition from "The rules prescribe such-and-such"
to "So I ought to do such-and-such" is to be legitimated by the
addition of a major premise to the effect that "We ought always
to do what the rules prescribe," the "ought" in the conclusion
has to be the same "ought" as the "ought" in the major premise.
Now the "ought" in the conclusion has its force only by virtue
of its use involving an appeal to the established rules; but this
feature must be missing from the "ought" in the newly supplied
major premise, since this "ought" cannot derive any force from
the very rules which it is being used to enjoin us to obey. Thus the
prescriptivist's attempt to bridge what he sees as a logical gap
fails. What was mistaken in his attempt from the outset was his
belief that there is a gap here waiting to be bridged. Why there is
not we can best understand by considering how "skyldr" ties
together "ought" and "owe." For the saga use of "skyldr" is
parallel to our use of "owe."
Whether I owe you money or not, how much I owe you, and
when the debt falls due for payment are questions of fact settled
by discovering the truth about antecedent transactions between
us, whether you lent me money and so forth. The institutional
framework of rules governing monetary transactions in general
and borrowing and lending in particular is such that given that I
received $5 under certain conditions I now owe you $5 ; just so
in the saga, given that the sons of Vestein killed Thorkell under
certain conditions, Bork now owes Thorkell's kinsmen their
death. Bork can of course ask: "Given that I ought to kill the sons of
Vestein, shall I do it?" But when all the facts are settled, he can-
not still ask: "But ought I to do it?" For that he ought to do it is
"Ought" 145
just one of the facts, in the only sense and use of "ought" available
to him.
But now the prescriptivist might retort that Bork's "ought"
has just been compared to "owe," and that although it may be a
question of fact whether I owe you $5, the question of whether I
ought to pay what I owe can always be raised, and that this ques-
tion is not a question of fact. But the prescriptivist's ability and
our ability to raise this question depends upon the availability
of an "ought" that is just not present in the language spoken by
Bork. What is required for that "ought" to be available is that a
new use be introduced, and that is to say, in this case, a new
social practice. Now it is clearly logically possible that at any time
someone may innovate radically. But the fact that this is always
logically possible does not entail that such an ability to innovate
was in fact present. Such an ability requires a detachment from
the established institutional rules which has yet to come on the
scene, which itself has certain social preconditions.
If the prescriptivist cannot make the "ought" expressed by
the saga's "skyldr" conform to his pattern of analysis, it is of
course also true that the naturalist cannot hope to succeed in this
project either. For the termination of the chain of justifications
in the sagas is simply the assertion of the relevant rules ; there is
no citing of some human good which will be procured by whatever
action is in question. Indeed, the fact that obedience to the rules
will produce disaster for a man is sometimes noted in the sagas,
not merely by the narrator, but by the agent himself. And this
contributes not at all to showing that the agent therefore ought
not to do what the rules prescribe. In a precisely similar way,
you may show me that if I pay my debts the results will be dis-
astrous for me or my creditors or both; but this does not show in
any way that I do not after all owe them money. Thus, this first
use of "ought" gives aid and comfort neither to prescriptivist nor
to naturalist.
It is of course open to either or both to say that this is a special
type of case, which we only recognize as morality and which we
only characterize by translating its key expressions with words
like "ought," because it can be and was as a matter of history
later on transformed into the kind of moral scheme which fits the
prescriptivist or naturalist pattern. But this, although perhaps
true, does not affect any of the preceding arguments.
Against the Self-Images of the Age
VI
Question XI: How we ought to aid our parents when in want ...
Question XVI: What reward is proposed by God for obedience
to parents ...
Question XX: With what punishments children are visited who
are neglectful of this commandment ... 1
VII
It is a banal commonplace that the breakdown of the theistic
framework produced a morally plural society; and in such a
society prescriptivism comes into its own. The reformed pre-
scriptivist whom I characterized earlier no longer insists that
anything can be a good or any principle a moral principle. In-
stead, he points out, when one has listed the recognized and
recognizable human goods one is faced with a large and indefinite
variety of competing alternatives. The admission that pain is an
evil is equally consistent with the morality of the sagas or the
Spartans, in which its point is to provide occasions for manifest-
ing fortitude and courage; the morality of the Christians in which
it is to be alleviated wherever possible; and the morality of de
Sade in which the infliction of pain is an expression of the agent's
liberty and autonomy. How does one choose between these
moralities? There is no way of taking the admitted list of human
goods and finding criteria for judging between such moralities;
for in the competition between moralities goods are matched
against goods, and there is no superior criterion of judgment.
In the modern world the sphere of morality is essentially the sphere
of tragedy as Hegel defined it. Consider, to take a quite different
type of example, some features of Kierkegaard's argument in
Either/Or.
In Either /Or Kierkegaard exhibited the necessity for a criterion-
less choice between two ways of life : the aesthetic, in which the
good of pleasure is allowed an overriding character, and the ethical,
in which the goods of a dutiful life are allowed an overriding
character. Suppose I try to find some criterion which will enable
me to decide between these two on impersonal grounds, some
criterion which will liberate me from the subjectivity of criterion-
less choice. If I ask: "Which life will satisfy my wants?" then I have
already in choosing this criterion chosen the aesthetic. If I ask:
"Which life ought I to pursue?" then the "ought" is the "ought"
of Duty, and I have already chosen the ethical life. Considerations
such as these led Kierkegaard to conclude that one can do no more
than offer rival descriptions of the two lives; the reader is then left
to choose, and the moral to be drawn is that no description can
possibly determine his choice.
His choice may of course be expressed simply by "~ shall choose
such-and-such"; but there is a modern use of "ought" according
6
Against the Self-Images of the Age
to which he might express a choice that he recommends to every-
one and which expresses his commitment to recommend it to
everyone by saying, "So I ought to choose so-and-so." This
"ought" is the "ought" of the prescriptivist analysis. In other
words, the argument that prescriptivism is always a correct
account of moral language because it is found to be a correct
account of these peculiarly modern situations of radical choice is
one that I have rejected throughout this essay; but this does not
entail that it is an incorrect account of these choice situations. For
in such situations ultimate moral principles do have to be chosen;
no description of the alternatives confronting us logically entails
any conclusion about what we ought to do or to choose, and
perhaps it is even the case that the only available criterion of
whether we have adopted one particular alternative is its embodi-
ment in our actions in the form of obedience to self-addressed im-
peratival injunctions. All that has been abandoned to naturalism
as regards this type of example is the claim that anything what-
soever could be a good.
VIII
IX
Two main theses have now emerged from my argument: the first
that naturalism and prescriptivism are most plausibly understood
not as rival accounts of the whole field of moral or even of evalua-
tive discourse, but as accounts of different types of moral and
evaluative discourse. These types of discourse I have tried to
discriminate in terms of three uses of "ought" which recur suc-
cessively in the history of that word in the English language and
English-speaking society. But I have also maintained that all three
uses of "ought" occur in current English and I want now to con-
sider this fact. Just as prescriptivism, in order that it could be
understood as a correct account of those fundamental moral
choices on which existentialist moral philosophers have laid so
much emphasis, had to be amended, so naturalism, in order that
it can be understood as a correct account of those situations in
"Ought" 1 55
I
There are two classical treatments of pleasure, the Benthamite 1
and the Aristotelian. 2 If I deal with them in that rather than in the
chronological order, it is because in recent philosophical writing
about pleasure, the Benthamite view has provided the target
for attack, the Aristotelian view the weapons. It seems clear
that Bentham treated both "pleasure" and "pain" as the names of
sensations. These sensations are distinct existences which may,
and on occasion do, exist without being accompanied by any act
of will. It is not entirely clear whether Bentham himself thought
the connection between pleasure and pain on the one hand, and
acts of will and actions on the other, to be purely contingent or
not. For one thing it is "the idea of pleasure" rather than the
sensation which has an effect on the will, according to Bentham.
But Bentham certainly appears to believe, first, that in all cases
where I take pleasure in or enjoy something, what constitutes the
pleasure or enjoyment is an accompanying sensation, which
exists over and above the object of pleasure or enjoyment; and,
second, that when I do something for pleasure or enjoyment and
achieve it, the pleasure or enjoyment is a sensation separately
identifiable from the means by which I procure the sensation.
Bentham gives fifty-eight synonyms for "pleasure"; his use of the
notion of synonymity reinforces the view that he takes there to be
a single, simple concept of pleasure, so that "enjoyment" for
example means precisely what "pleasure" means and names
precisely what "pleasure" names.
In Aristotle's view, which is far less tidy than Bentham's,
pleasure is analysed at one point in terms of unimpeded activity
and at another in terms of its resemblances and differences to an
1 J. Bentham, Principles of Morals and Legislation (New York: Hafner,
1948), Chapter 4 and elsewhere.
2 Aristotle, Nicomachean Ethics (London and New York: Heinemann
II
First of all, then, to remark some of the variety of concepts:
x. "Heureux qui, comme Ulysse ... " (du Bellay); "0 happy is
the man who hears instruction's warning voice" (Ps. i, Scots
metrical version). The use of "happy" to express a verdict on a
man's life depends upon a sense of "happy" in which a man may
truly be called happy who has suffered a great deal, as well as a man
who has prospered.
2. I may be called happy in the above sense, because during my
life I have enjoyed myself on many occasions. But, in giving this
supporting reason for the verdict, I am using "enjoy" to mean
something other than "be happy." I can enjoy a game, a holiday,
Pleasureas a reasonfor action 1 77
a friend's company, digging, editing a Greek text. My motive in
doing these things need not be a wish for enjoyment for me to
enjoy whatever it is; I may play games or dig for the sake of my
health and edit a text because I need to earn money, and yet
enjoy what I do. Whether I enjoy what I do or not is a question
that others may answer by observing whether or not I try to
prolong the activity, appear absorbed in it, yawn, and so on. But
my own testimony is highly important, although not necessarily
always conclusive.
3. An activity, a sensation, a sight or sound, a work of art, a
taste, a smell may be called pleasant or ·unpleasant, enjoyable or
unenjoyable. In so characterizing it, I do not commit myself to
having enjoyed it or found it pleasant on any particular occasion.
It may well be that it takes experience to find it pleasant, or it
equally may well be that familiarity dulls the pleasure. " ... 'I add
to them, in the laying out of grounds, a third and distinct character,
which I call unexpectedness.' 'Pray, sir,' said Mr. Milestone,
'by what name do you distinguish this character when a person
walks around the grounds for the second time?' " (Peacock).
Moreover, when I call something pleasant or enjoyable, although
I may not be able to give reasons for my characterization, it
always makes sense to ask for them. I may call a drink pleasant
because it is refreshing, a holiday enjoyable because it combined
sun and sea with the opportunity to look at paintings. If I call
something pleasant or enjoyable I am saying, or at least giving
my hearers to understand, that the standard criteria for that sort
of thing's being an enjoyable one have been satisfied, at least where
there are such criteria. And commonly there are.
4. Pleasure as a distinct object of pursuit might be said to
consist in the enjoyable qualities of those activities, sensations,
tastes, and the like, which are, and can only be, sought for the
sake of the pleasure found in them. And certainly even if I not only
enjoy nursing the sick or working on an automobile production
line, but do these things because I enjoy them, I cannot be said to be
devoting myself to pleasure in this sense. Yet to say this would be to
attend to one end of a spectrum only, a spectrum which must be de-
fined partly in terms of this end of the scale and partly in terms of
the contrast between activities which can only be carried on for the
sake of pleasure and activities which could in no sense be said to be
pursued only for pleasure's sake, even when we enjoy them. For
there are many activities which carried on in one type of context
could only be done for pleasure, but which in another type of
context could be carried on for other reasons. Such is fishing. And
Against the Self-Images of the Age
there are activities which embody values so central to human life
that although they are highly productive of pleasure, they could not
be undertaken without some attention to those values and so could
not be done purely for pleasure. Such is mountaineering. But there
are at the opposite end of the spectrum from that of which I first
spoke, when the activities to which I wanted to draw attention
were eating and drinking what is not designed to prevent hunger
and thirst, the enjoyment of tastes and smells, sensual pleasure, and
all those items to which the OED may be taken to be referring
when it gives as its second definition of "Pleasure": "In bad sense:
Sensuous enjoyment as a chief object of life or end in itself."
The most important ground on which I have distinguished
between these four concepts is the difference in each case in what
would make statements embodying them true or false. This
attention to truth-conditions, rather than to syntactical distinctions
or I idiomatic nuances, is necessary because the words used to
express these concepts are, in contemporary English as contrasted
with the English of earlier generations, often interchangeable.
There are indeed idiomatic points to be made of philosophical
interest: tastes, smells, and sensations are usually said to be
pleasant or pleasurable, activities to be enjoyable. In this way
different parts of the spectrum of pleasure are distinguished in
ordinary usage. But for most idioms that make use of some form
of "enjoy" and its cognates there is an equivalent idiom using
"please" or "pleasure." It is because of these facts that it would be
equally wrong either to assert with Bentham that "pleasure" and
"enjoyment" are synonyms or simply to deny Bentham's assertion.
Hence also the misleading character of all such assertions is that
pleasure is a species of enjoyment. For such assertions presuppose
what is not the case, that the vocabulary of pleasure and enjoyment
is a currency with fixed values. In my own elucidation I have
annexed certain words for certain concepts in an arbitrary but,
I hope, tolerably lucid way.
Some of the relations between the four concepts I have sketched
are fairly clear: that a man has enjoyed much of his life is a good
reason, although it may not be a sufficient one, for calling him
happy; that an experience or activity was pleasurable may be
cited as an explanation for having enjoyed it. But others of the
relationships are more complex. The one on which I wish to lay
stress is that between first-person reports of what I do or did take
pleasure in or enjoy and statements about an experience or
activity being enjoyable, pleasant, or pleasurable. When asked
why I enjoyed an experience or why the experience was enjoyable
Pleasureas a reasonfor action 179
I may in both cases point to features of the experience which
made it enjoyable. But if I fail to enjoy an experience and am
asked to explain why I so failt;_d,I may do one of two things. I
may say that the experience just was not enjoyable, and I may
cite features of the experience which made it unpleasant. Or I may
account for my lack of enjoyment alternatively by explaining
that although the experience had all the features of an enjoyable
experience, none the less I failed to enjoy it because something was
wrong with me: I had a cold, I was overtired, or even-the
residual category-I just did not feel like it.
I lay stress on these alternative directions in which explanation
may move (in calling them alternatives I do not mean to imply
that it could not be the case both that the experience had un-
pleasant features and that I was in no fit state to enjoy it), both
because this contrast between pointing to the state of the agent
and pointing to the features of his experience or activity will
recur later in the argument, and because attention to it enables us to
diagnose certain errors. Ryle opens a discussion of pleasure by
asking what sort of difference the difference is between a walk which
one enjoys and a walk which one does not enjoy. The point of his
question is partly to bring out that the difference is not that one
walk is accompanied by a certain specific sensation and that the
other is not. With this no one but a Benthamite could quarrel. But
the question, as Ryle poses it, is almost as misleading as a Bentha-
mite answer to it would be; for the form of the question suggests
that one is looking for a single answer. And this is the kind of
answer Ryle gives. A walk that one enjoys is one which has
absorbed one's attention. So Ryle has written:
To say that a person has been enjoying digging ... is to say that
he dug with his whole heart in his task; i.e., that he dug,
wanting to dig, and not wanting to do anything else (or nothing)
instead. His digging was a propensity fulfilment. His digging
was his pleasure, and not a vehicle of his pleasure. 1
This passage contains a mixture of true and false, When Ryle
explains enjoying digging in terms of wanting to dig he gives us
an important clue to a correct analysis. But he presents this clue
as though there were only one answer to such questions as: "Why
did you enjoy digging today?" or "What is the difference between
that walk yesterday which you enjoyed and the one today which
1 Gilbert Ryle, "Pleasure," Proceedings of the Aristotelian Society,
Supplement, 1954 and The Concept of Mind (London: Hutchinson; New
York: Barnes & Noble, 1949), p. 108.
180 Against the Self-Images of the Age
you did not enjoy?" In fact the moment it is clear that we are
talking about digging or walking on a specific occasion it also
becomes clear that our enjoyment or lack of it has to be connected
with features specific to the occasion. The difference between the
walk today and the walk yesterday may be the difference between
a walk when the sun was shining and a walk when it was cold and
wet or the difference between a walk with Sarah who is charming
and adores you and a walk with Selina who is boring and dislikes
you. Suppose that it is objected that this is not the type of case
Ryle had in mind. For a walk with Sarah is not the same activity
as a walk with Selina, and a walk in the sun is not the same activity
as a walk in the rain. What about the case where you enjoy and
fail to enjoy the same activity on two successive occasions ?
The answer can only be that in that case the answer must be
looked for in your state of body or mind and not in the activity.
Again there will be no one general answer. You may on one
occasion have had a cold or been overtired, or on the other have
just had good news or felt unusually well.
So that we have to pass from the walking or the digging to the
enjoyment by way of features of the walking or the digging. What
makes these features relevant to cite is the fact that they are of a
kind recognized as making a walk or a dig pleasurable. "I enjoyed
the walk because the sun was shining" is intelligible as it stands ;
"I enjoyed the walk because of the cold drizzle" is not. Of course
when someone says "I enjoyed ... " his assertion does not mean
the same as an assertion about what made the occasion enjoyable.
What his assertion does mean, however, will not be understood if,
as Ryle does, in the earlier of the discussions cited, we try to
connect the enjoyment solely with the agent's mode of activity.
Professor W. B. Gallie,1 and Mr. C. C. W. Taylor 2 have both,
for example, taken pleasure to be some form, mode, or species of
attention. A general view of their case has been given by Professor
H. L. A. Hart 3 who has written approvingly that,
the outlines at least of a new and more realistic [than the
Benthamite] analysis are clear. The elements previously treated
as mere empirical evidence of a separately identifiable sensation
of pleasure have now been introduced into the analysis of
pleasure. The wish for prolongation of the activity or experience
1 W. B. Gallie, "Pleasure," Proceedings of the Aristotelian Society,
Supplement, 1954.
2 C. C. W. Taylor, "Pleasure", Analysis, Supplement, 1962.
3 H. L.A. Hart, "Bentham," Proceedings of the British Academy, 1962.
Pleasureas a reasonfor action
enjoyed; the resistance to interruption; the absorbed or rapt
attention; the absence of some further end beyond the activity
enjoyed-these are surely conceptually and not merely
empirically linked with pleasure.
III
(London and New York: Oxford University Press, 1963), pp. 125-29.
I would not want the case on pleasure, which I have constructed in
outline from Hare's views on pain, to be treated as though Hare could be
held responsible for it.
3 K. Baier, The Moral Point of View (Ithaca, N.Y.: Cornell University
own exposition.
186 Against the Self-Images of the Age
as objects of desire or aversion. If there are such predicates and
properties, their existence presupposes that certain desires and
aversions are standard for human beings, standard not only in
being statistically usual, but in providing norms for desire, to such
an extent that our descriptive vocabulary embodies these norms.
Hence certain types of statements would furnish us with reasons
for action in a special sense. Hare's view on pain might be genera-
lized into the contention that there are no such statements.
My first problem about Hare's thesis is equally a problem about
the statements of Ryle and Baier from which Hare's exposition
started. All speak of sensations which we dislike. Now certainly
we dislike many sensations, and pains are among these; but the
variations in my dislike of my pain are not to be confused with
variations in the pain. That is, it is just not the case that the more
painful my pain the more I dislike it. For my dislike may intensify
or lessen depending on how well able I feel to bear the pain; and
there are times when I may feel better able to bear a greater pain
than I am able to bear with equanimity at other times. Equally,
my dislike of the same pain of unchanging intensity may vary
greatly. Thus, the connection between pain and characteristic
pain behavior is unlikely to be elucidated at all by putting this
concept of liking or disliking in the center of the picture.
Consider instead the notion of "the same sensation." What
are the criteria for identity or similarity of sensation? Certainly
identity of stimulus is not enough. It always makes sense to inquire
whether the same stimulus is producing the same sensation as it
did before, or whether it is producing the same sensation in you
as it is in me. Equally, reactions to sensations vary. It makes sense
to inquire whether the same sensation produces the same or
different reactions in me as it does you, or in me at different times.
But what criterion have we for characterizing sensations at all,
if this is so? It is considerations such as this that lead either to
skepticism or to behaviorism. The behaviorist attempt to outlaw
the sensation itself breaks down, among many other reasons,
because I may, when I cannot observe a stimulus, infer correctly
what it is from a sensation (as when I infer that someone has
stabbed me in the back). But the comparison of one sensation with
another-which is presupposed, for example, in such correct
inferences to true conclusions-is only possible because a relatively
uniform intervening sensation between stimulus and behavioral
response is presupposed. Take the notions of tickles and itches.
We have to acquire these notions in terms both of their charac-
teristic causes and of the characteristic responses to such causes.
Pleasure as a reason for action 187
If tickling in certain parts of the body did not characteristically
produce laughter, we should lack the concept of a tickle. Again,
if a feeling did not characteristically produce an impulse to
scratch we should hesitate to call it an itch.
When I speak of a behavioral response, I mean a natural,
primary response; I may always learn to inhibit such responses,
and there is no logical limit to such learning. So the notion of an
itch that I do not any longer want to scratch or a tickle that pro-
duces no laughter is perfectly intelligible. So is the notion of a pain
that produces neither clutching nor screaming. But is it then true
that having used the behavioral concomitants to acquire a sensa-
tion vocabulary, I can then use the vocabulary without any
behavioral reference? Consider the notion of severe and less
severe pain. What can this be but pain to which the natural response
would be more and less extreme forms of avoidance or aversion
behavior? Or, if one prefers, pain which requires more or less in-
hibition of such behavior. Equally, the same pain must be pain
whose natural response would be avoidance behavior of the same
degree, although it must also be pain of the same kind-stabbing,
throbbing, or aching, for example. What otherwise could "the
same pain" mean? Suppose the reply to this is: the criterion of
identity is just that the two feel the same, they are phenomenolo-
gically identical. Can we make sense of the notion of two sensations
feeling the same but being associated with different natural
behavioral responses? If this were so, ascriptions of sameness of
sensation in different people would not be possible. Predicates
ascribing sensation would be egocentric predicates, which they
are not in fact. Of course, we have the concept of the phenomeno-
logical feel of sensations; but the vocabulary in which we express
it depends on a notion of sensations as comparable, which depends
in turn on the association of sensation with behavioral responses.
This dependence is neither purely contingent, as Hare supposes,
nor purely analytic. In fact, half our difficulties have arisen from
too rigid an application of an analytic/synthetic dichotomy.
Consider a parallel case, that of fear. We certainly have feelings of
fear, which sometimes simply arise in us inexplicably and which
we recognize by their phenomenological feel, although we may
know perfectly well that at that moment there is nothing to be
afraid of. Should we therefore conclude that the connection be-
tween feelings of fear and the belief that something harmful or
dangerous is at hand, which occurs in the majority of cases of fear,
is a purely contingent connection? This absurd conclusion is
avoided by distinguishing between primary cases of fear and
188 Against the Self-Images of the Age
secondary cases. We are only able to use the concept in the secon-
dary cases because we understand it in its primary application.
So also with pain. There may be cases where we wish to assert
an identity of sensation, sufficient to call the sensation "pain"
between the central cases where the notion of pain is conceptually,
though not analytically, tied to the notion of avoidance behavior
and the marginal cases where learned inhibition or physiological
interference has broken the link between the felt sensation and the
behavior. If we say this, we are of course committed to disagreeing
with Hare's view that we teach and learn the word "pain" by
means of a guess based on the presence of a purely contingent
criterion of pain, the external behavior. That Hare has to use
the word "guess" to express his view is important; if he were
right, there would always be room for doubt as to whether I had
learned to understand the word correctly. But this doubt would
apply to everybody, and where such a doubt applied there would
be no sense to the notion of a correct understanding of the word.
That is, the consequence to be drawn from Hare's view would in
fact be that we had no clear concept of pain at all.
The view that I have outlined allows of course for the occur-
rence of cases of many kinds where pain is not accompanied by
avoidance behavior; some of these cases present us with no diffi-
culty at all, others such as the lobotomy cases are genuinely hard
to understand. But it is crucial to note that the difficulty in under-
standing them does not arise from any philosophical theory about
the meaning of "pain" and kindred words but from the language
itself. The ordinary speaker is as puzzled as the philosopher by
the avowals of such patients. The same is true of masochism.
The relevance of all this to pleasure has now to be brought out.
It is, I hope, clear that there is no logical barrier to the existence
of a vocabulary of pleasurable sensations, in which the identifica-
tion of the sensation is tied to a certain type of behavioral response,
so that where the response was lacking we should have at least a
strong ground for inquiring whether the sensation could be the
same. What is said of sensations would apply equally to tastes and
smells. Let us apply the account I have given of pain to a parallel
case of pleasure. If I put my hand in the fire and let it roast there,
it can be explained either that I have trained myself to be heroi-
cally stoical in the face of pain and have some good reason for my
action or that I am physiologically abnormal or anesthetized. What
cannot be said is that I feel the pain just as anyone else would, am
neither stoical nor abnormal or anesthetized, but just do not mind
the pain. For my behaviour is a sufficient reason for concluding
Pleasureas a reasonfor action 189
that if I am not being stoical, then, whatever I feel, it is not in any
possible sense pain. Equally, if I when thirsty drink cold water
from a mountain stream, and then spit it out in disgust, it may be
explained that the water is polluted or that my mouth or throat is
in an abnormal condition. But what cannot be said is that there is
nothing wrong with the water and that my mouth and throat are in
a normal condition, but that I just find the taste of cold water,
which most people like, intolerable. For that someone finds such
a taste intolerable is a sufficient reason for concluding that the
taste cannot be the same taste that the rest of us experience. If
someone finds sexual sensations not pleasurable but painful, he
does not remark to himself that his likes and dislikes are those of a
minority; he looks for a physiological cause of his abnormality.
That is to say, we do in fact treat certain tastes and sensations
as pleasurable as such. If someone does not take pleasure in these,
we look for an explanation of his failure in terms of the state of
his body or his mind. The taste of cold water is not especially
pleasurable perhaps; the tastes of Guinness or Chateau Yquern
can be liked or disliked; we allow a wide range of variation to
taste. But there is a limit to this range beyond which we explain
lack of pleasure as we do lack of pain, in terms of the subject's
discrepancy from norms of desire, norms which are embodied in
parts of our vocabulary of pleasure and pain.
When I report what I enjoy, I am, unless there is reason to
believe that I am insincere, the final authority on the truth of my
report. When I report that an experience was or was not enjoyable
or a sensation pleasurable, I can without insincerity be saying
what is false. For the standard of the enjoyable and the pleasurable
is not private, but public. De gustibus est disputandum. To call
something pleasurable, therefore, is partly to say that it embodies
the object of desire from the standpoint of the norms of desire.
If I tell you that there is a fun-fair at the end of the pier I only give
you a reason for going to the end of the pier if you happen to like
fun-fairs and want to enjoy one now. If I tell you that fun-fairs
are pleasurable, I may give you reason to believe that you would
like them. Thus, statements about what is pleasurable do afford
us in a special sense reasons for action. So the statement which
I quoted from Dr. Kenny at the outset turns out to be true. It is
also of course true that the prospect of pleasure will not move
me to action unless I want pleasure at the given moment. But to
ask why I want pleasure is indeed, as Dr. Kenny suggested,
wrong-headed; because the notion of pleasure is the notion of a
property of certain activities and experiences which are treated as
Against the Self-Images of the Age
standard objects of desire, which help to define not merely the
desired, but the desirable. And to ask why I want what is desirable
would only have point if I was thought to be in some way abnormal
or perverse. That is, it may on occasion be asked why I should
want pleasure. Moreover there are other desirable objects as well
as pleasure; so that it makes sense to ask why I do not want pleasure
at this moment. But it is after all not therefore true that, if a man
wants pleasure, one can intelligently ask why. It is pleasant to dis-
cover that this obvious truth is an obvious truth after all.
The antecedents of action 1
I
We are haunted by the ghosts of dead concepts. The trouble with
ghosts is that they do not replace the living satisfactorily and yet
do not leave us with an entirely vacant hearth either. One such
dead concept is the concept of the will; its ghost is the philo-
sophical theory that the line which can be drawn between what is
a human action and what is a mere happening is such that actions
cannot have causes in the way that happenings can. When I speak
of the concept of the will I do not, of course, refer to pellucid
colloquialisms as in "Where there's a will there's a way" or "a
strong will"; I refer to the concept built up in post-medieval
philosophical psychology-in Hobbes, in Hume, and in Kant, for
example.
The exercise of the will in Hobbes distinguishes human action
from animal behavior because it presupposes a capacity for
deliberation. "In deliberation, the last Appetite, or Aversion,
immediately adhering to the action, or to the omission thereof, is
what we call the WILL; the Act (not the faculty) of Willing." 2
The exercise of the will in Hume distinguishes human action
from muscular or nervous responses because it involves con-
sciousness. "I desire it may be observed, that, by the will, I mean
nothing but the internal impression we feel, and are conscious of,
when we knowingly give rise to any new motion of our body, or
new perception of our mind." 3 The exercise of the will in Kant
marks out the human action from mere physical movement by
making action movement in accordance with and in obedience to
1 Reprinted from British Analytical Philosophy, edited by Bernard
Williams and Alan Montefiore (London: Routledge & Kegan Paul;
New York: Humanities Press, 1966).
2 Leviathan, I, 6.
3 Treatise, II, iii, 1.
191
Against the Self-Images of the Age
precepts or rules. "Everything in nature works according to
laws. Rational beings alone have the faculty of acting ... accord-
ing to principles, i.e., have a will." 1
The concept of acts of will which emerges from these quotations
is one according to which the will is a special kind of efficient cause,
the necessary cause of any human action. To make an act of will is
to make a conscious and rational decision. It is to embody a pre-
cept for action in an instruction to oneself. Saying to oneself "So
I will do such-and-such" sets one's limbs in motion. On occasion
one may fail to set one's limbs in motion, just as any other cause
may fail to operate if prior causes intervene. One's limbs are
paralysed or shot away. The first requirement in explaining an
action therefore is to assign a proximate cause to the action by
pointing to a prior act of will. In Hobbes and Hume appetite or
aversion inspire and inform such acts; in Kant the causal chains
which terminate in inclination may always fail to operate because of
the prior intervention of that uncaused cause, the autonomous
rational will obeying its self-imposed categorical imperative.
This ancestry makes it less surprising that the concept of acts of
will was later called upon to play opposite parts by different
philosophers. In the mechanistic psychology which the utilitarians
took over from Hartley all human actions are the determinate
effects of prior causes, and in the causal chain the act of will is the
immediate cause of the action. To some anti-determinist writers
the will is the intervening cause which prevents human action being
the mere outcome of events in the brain or the nervous system.
So participants of both determinism and free will invoke "the
will." For both parties, acts of will possess two characteristics
which are used by later writers to attack their existence: they are
events distinguishable from actions, always as a matter of contin-
gent fact preceding them; and they are events necessarily con-
nected with actions in that without them what followed would
not be an action. It was in this way that H. A. Prichard, for ex-
ample, wrote of acts of will. And it is in this way that his critics
have written of them in order to cast doubt on their existence.
But the point at which genuinely sharp criticism of the concept of
the will began was not here; it was the dualism which the concept
implied that first attracted hostile critical attention.
II
The doctrine of acts of will from Hume to Prichard was formu-
lated by philosophers who accepted a dualist view of body and
mind, and to this extent were true children of Descartes. This
dualism may have been refuted by Hegel, but in England, until
recently, Hegel and mystification were almost synonymous. The
refutation of Cartesian dualism was therefore in England the work
not of Hegel but of Professor Gilbert Ryle, a chapter of whose
The Concept of Mind is explicitly devoted to the will, but whose
argument throughout the book is extremely relevant.
The central argument of The Concept of Mind is that the criteria
for the application of those expressions which we use to describe
mental activity are all criteria of success or failure in performance
in the realm of overt behavior, and that therefore we neither
need nor have reason to postulate a realm of specifically mental
acts above and behind such behavior. Foremost among the
reasons which have misled philosophers into supposing that there
are such mental acts is a false view that those bodily movements
which are to count as human actions must have a special sort of
mental cause. The application of this doctrine to what I shall now
call the traditional view of acts of will is obvious.
Ryle in The Concept of Mind does not (at least nor usually) want
to deny the occurrence of any of the familiar "inner" events, such
as twinges or pains at one end of the scale or musings and interior
monologues at the other. What he does want to deny is that these
could have the characteristics which mental acts are alleged to have
in the traditional doctrine. In the case of acts of will, Ryle argues
that we cannot identify such acts (which he calls volitions) with
"such other familiar processes as that of resolving or making up
our minds to do something" or setting ourselves to do something.
For we know that there are many human actions which do not, in
fact, follow on such familiar processes, without thereby ceasing
to be human actions. But in the traditional doctrine any action
springs from an act of will. Hence these familiar processes and
events cannot be what the traditional doctrine wished to identify
as such acts.
Moreover, and here we return to Ryle's central argument, when
we describe actions by using such characteristic predicates as
"voluntary" or "responsible" or "done on purpose" or when we
insist that such-and-such a movement was not an action ("He was
1 94 Against the Self-Images of the Age
pushed," "He slipped"), we seek to establish the truth of our
description by reference to properties of the overt performance.
We never deem it logically appropriate to inquire as to the presence
or absence of acts of will. But if the traditional doctrine were
correct, this would be the appropriate and the only appropriate
question.
Ryle himself seems to place great weight on another line of
argument with which it is less easy to be happy.
No one ever says such things as that at 10 a.m. he was occupied
in willing this or that, or that he performed five quick and easy
volitions and two slow and difficult volitions between midday
and lunch-time .... Novelists describe the actions, remarks,
gestures and grimaces, the daydreams, deliberations, qualms
and embarrassments of their characters; but they never mention
their volitions. They would not know what to say about them.
This appeal to what "no one ever says" or to what everybody does
say is in itself ambiguous. It may simply be a way of underlining
the point that actions can be adequately characterized in all
possible ways without bringing in the notion of acts of will and
that the occurrence of such acts is therefore an unnecessary hypo-
thesis. But it suggests something else, in the form in which Ryle
advances it; namely, treating ordinary nonphilosophical modes
of speech as canonical for philosophical analysis.
If this is the thesis, it may once again be construed in two ways.
A weak and unobjectionable version of the thesis is simply that
any distinctions marked in ordinary language are likely to point to
differences which philosophers ignore at their peril. But there is
a stronger version of the thesis which must appear much more
disputable. This is the thesis that "ordinary language is in order,
just as it is" and that in the elucidation of what human action is,
common speech is not merely a source of suggested distinctions,
but provides us with hard criteria.
A quite different type of argument, which Ryle has used against
what he takes to be mythological mental acts, has been advanced by
A. I. Melden, specifically against the occurrence of volitions. 1
Melden argues that the concept of a volition involves an infinite
and vicious regress. For, on the traditional view, to move my limbs
I must first perform an act of will. But an act of will is itself an
action which I peform. And every action has to be preceded by an
act of will. So the performance of an act of will must itself be
preceded by an act of will, and so proceed ad infinitum. So an
1 Free Action (London: Routledge and Kegan Paul, 1961), Ch. 5.
The antecedentsof action 195
III
The act of will was presented as the cause of the human action.
But if there are no acts of will, as Ryle and Melden argue, do
actions lack causes? Or have they quite other causes than acts of
will? The discussion which has followed on from attempts to
answer these questions can only be fruitful if we distinguish care-
fully between three senses of "cause," or at least between three
ways in which causal questions can arise. There is first of all what
is usually spoken of as Humean causality. This is the view of
causality which springs from one of Hume's several and incom-
patible accounts and which was further developed by J. S. Mill.
In this view one event is the cause of another, if and only if events
of the former type have uniformly been observed to precede
events of the latter type, and events of the latter type have uni-
formly been observed to follow events of the former type. The
occurrence of the earlier event is both a necessary and a sufficient
condition of the occurrence of the later event.
It is this concept of causality whose application has aroused
controversy over determinism. If actions are the determined
outcome of prior events, and presumably of prior physiological
events, it has seemed difficult to draw a distinct line between an
action and a mere reflex, and certainly difficult to draw the kind of
distinction which would lead us to impute responsibility in one
case and not in the other. It is, perhaps, because overtones of the
determinist controversy lie in the background that discussions of
the causality of actions have been directed so overwhelmingly
toward Humean causality. But, in fact, no discussion could be
carried to a successful conclusion unless it attended to at least two
other senses or analyses.
One of these is the sense of "cause" which is equivalent to
"necessary, but not necessarily sufficient, condition," a sense which
Against the Self-Images of the Age
is apparently rather than really simple. For very often when we
speak of "the" cause of an event, for instance at a coroner's court
in assigning responsibility for an accident, we point to a condition,
by itself necessary but not sufficient for the occurrence of the
accident. We do so when events were in train such that without
the condition in question being satisfied the event would not have
occurred. Tak.en by itself the condition was necessary but not
sufficient. Tak.en in conjunction with all the other prior events, its
satisfaction was sufficient to bring about the accident. So it is with
the ice patch on the otherwise safe road. The point to note here is
that what is by itself only a necessary condition for the occurrence
of an event can be used to bring about the event, or can be re-
ferred to in giving a causal explanation of the event provided only
that we know when its addition to other conditions is sufficient to
produce the event. But by referring to such an occurrence as a
cause we do not commit ourselves either to a generalization of the
form "Whenever ice patches occur there is an accident" or to one
of the form "There is never an accident unless there is an ice-
patch," but only to one of the form "Whenever such-and-such
other conditions occur, then, if there is an ice patch, there will be
an accident." The importance of generalizations of this type needs
much more attention, but for the moment we should only note
that the task of detecting necessary conditions as it leads up to this
type of generalization is inseparable from the task of detecting
sufficient conditions and thus of formulating generalizations of
the Humean type.
The third sense of "cause," or the third point about the sense
of "cause," is one that is not incompatible with but required by the
other two. This is the sense which was underlined by Professor
H. L.A. Hart and Mr. A. M. Honore in Causation and the Law.
Here a cause is a lever, a means of producing some other event.
There could be no well-established Humean generalizations
unless we were able to interfere with the course of nature and so
discover whether apparently uniform sequences were genuine ones
or not. But the Hart-Honore analysis brings out the importance
for causality of the concept of what would have happened if the
cause had not operated. All causal explanation presupposes a
background of generalizations about what occurs in the absence
of the cause. This is true of both cause understood in Hume's
sense and of cause understood as necessary condition.
The antecedents of action 197
IV
This very inadequate sketch of causality is a necessary prelude to
examining the two main attempts to show that actions cannot
have causes, at least in the Humean sense. The first of these derives
from an attempt to correct the assumption that the necessary and
sufficient conditions of human action are to be found in prior
physical events, an assumption which depends upon a companion
assumption that human actions are in fact only extremely complex
physical movements.
Against whom is this insistence directed? The answer is that a
great deal of physiology and psychology has taken it for granted
that this is correct. All attempts to explain human action by building
cybernetic models assume that human actions are of the same
kind as the movements of such models. The greatest of the
behavior theorists of modern psychology, Tolman and Hull, set
themselves the explicit goal of explaining human actions as very
complex exemplifications of fundamentally simple patterns of
physical movement. Those philosophers who have tried to show
the falsity of this have clearly wanted a concept or set of concepts
which will perform the function that the traditional concept
of the will performed. But they have moved in a quite different
direction.
In an article which tries to show that Hobbes and Hull were
both essentially pursuing the same goal of mechanical explana-
tions of human action, R. S. Peters and H. Tajfel have pointed
out that bodily movements cannot be the genus of which human
actions are a species, because the same bodily movements can be
used in performing quite different actions and the same action
can be performed by means of quite different bodily movements.
So these bodily movements which are employed in writing a man's
name may be used in signing a check, or giving an autograph, or
authorizing a representative. Equally, the same action of paying a
debt may be performed by those bodily movements involved in
signing a check or by those involved in handing over coin. In
other words, the criteria which we imply in judging that two bodily
movements are the same or different are quite other than the
criteria which we use in judging that two actions are the same or
different.
Actions then cannot be identified with bodily movements. But
while in the traditional view actions were bodily movements plus
Against the Self-Images of the Age
something else-namely, an act of will-for the more recent view
this is equally incorrect and misleading. For to speak of human
actions is to speak at a different logical level from that at which we
speak of bodily movements. To call something an action is to
invite the application of a quite different set of predicates from
that which we invite if we call something a bodily movement. If I
say "I moved my arm," I do not say either the same or more than I
say if I say "My arm moved." I bring what occurred under a
different form of description. We can bring out this difference in a
number of ways. First, if we ask "Why did your arm move?" we
invite a causal answer including perhaps a story about conditioned
reflexes and a story about muscles and nerves. If we ask "Why did
you move your arm?" we invite a story about intentions and
purposes. Equally, if on being asked to explain a piece of be-
havior I start to give an account in terms of muscular and nervous
mechanisms, I thereby treat the behavior as a piece of physical
movement and not as an action. If, on the other hand, I talk about
purposes, goals, desires, intentions, or the like, I thereby treat the
behavior as an action. Secondly, if I say "I moved my arm," then
the question "What reason did you have for doing that?" is
always in place, even if the answer is, "I do not know why I did it."
To say "I do not know" here is not to say "There is a reason, but
I am ignorant of it" (except in psychoanalytic contexts, which
demand special treatment). It is to say in effect "I had no reason,
though I might have had." And nobody can know my reasons or
lack of them, unless I tell or otherwise betray them. J-lere I have
special and unique authority. But when it is a case where it is
appropriate to say "My arm just moved," I have no such special
authority in giving explanations. What is needed is not the
authority of the agent as to his own intentions and purposes but
the authority of the physiologist on matters concerning condi-
tioned reflexes, nerves, and muscles.
Thirdly, in the standard cases at least where I say "I moved my
arm," there is no room for the question "How do you know?" The
reason for this can be brought out as follows. The point of asking
"How do you know?" is to ask for the credentials of a claimant to
knowledge in cases where the claimant to knowledge may be in
either a better or worse position to back up his claim. "Pegasus
won the 3.30." "How do you know?" "I was on the course," "I
saw it on television," "My bookmaker told me," "I saw it in a
dream" are all possible answers-of quite different value. But
where self-knowledge of my own present actions is concerned,
there is no question of being in a better or worse position to know.
The antecedents of action 1 99
And so there is no room for the question, "How do you know what
you are doing?" But there is room for the question "How do you
know?" where not actions but bodily movements are the subject
of the inquiry. Usually, of course, the answer is very simple:
"How do you know your arm moved?" "I felt it move," But from
a partially anesthetized man, lying so that he could not see his arm,
the answer "I saw it in the mirror" would make sense. And so
would any other answer which appealed to observation or inference.
Whereas this type of answer would make no sense as a reply to
the question: "How do you know that you moved your arm?"
(Indeed, this question, as I have already pointed out, lacks point
except perhaps in some very special contexts. It does not follow
that a man may not say "I moved my arm" outside those contexts
and be mistaken. Where the question "How do you know?" lacks
application there is still room for error.)
Fourthly, in cases where it is appropriate to say "I moved my
arm" rather than "My arm moved" the future tense used before
the event would express an intention, not a prediction, if used in
the first person. Moreover, if it is appropriate to say of the event
afterward "I moved my arm," then neither I nor anyone else
could predict that I would move my arm except on the basis of a
knowledge of my intentions. It does not however follow, as has
sometimes been argued, that where an event is the object of my
intention it cannot be the object of my prediction. What does
follow is that the expression of my intention is never the expression
of my prediction. But the expression of an intention and the ex-
pression of a prediction can be closely related; for if I express an
intention, I license others to predict. Of course, whether they are
wise to predict or not depends upon the evidence they possess as
to how far I am usually faithful to my declared intentions. Now
their beliefs (or their knowledge) on this point will be expressible
upon occasion in the form of Humean generalizations. Moreover,
I can acquire such knowledge about myself. A man may come to
recognize his own reliability or unreliability. Consequently, even
if it is psychologically out of the way, it is not conceptually odd
for a man to say "I fully intend to do it tomorrow, but I know how
unreliable I am, and so perhaps you are right and I will fail to do
it again tomorrow." What is more, in the very framing of his
intentions a man's self-knowledge and predictions about his own
reliability inevitably come into play. Hence, in cases where it is
appropriate to say "I shall move my arm," prediction is dependent
upon knowledge of intention, but intention need not be entirely
divorced from prediction ; nevertheless, prediction depends not
200 Against the Self-Images of the Age
at all on knowledge of intention where it is in place to say "My
arm will move."
Fifthly, the point of distinguishing between "My arm moved"
and "I moved my arm" is brought out very clearly in just those
borderline cases where we are uncertain which to say. We are all
familiar-from novels if not from experience-with cases where,
as we say,' the body seems to have taken control. In Sartre's
novel L' Age de Raison the hero-no, the protagonist-Mathieu,
intends to say to his mistress "I love you" and finds himself saying
"I don't love you." Do we describe this as something he said or as
words (or perhaps sounds) that come out of his mouth? Is it action
or bodily movement? How we answer the question in this parti-
cular context does not matter for our present purposes. What does
matter is that we cannot evade asking it, that we cannot escape
the distinction between action and bodily movement.
Yet what follows about causality? Only that if we are to look
for the causes of human actions, then we shall be in conceptual
error if we look in the direction of the causes of the physical
movements involved in the performance of the actions. It does not
follow that there is no direction in which it would be fruitful to
search for antecedent events which might function as causes.
What has suggested this further conclusion is the way in which
the investigation of concepts very close to the concept of action,
such as that of intention, has been carried through. Wittgenstein
wrote: " 'I am not ashamed of what I did then, but of the inten-
tion which I had.' 1 And didn't the intention reside also in what I
did. . . .'' 1 Just because the intention resides in the action, it
comes too close to it to play a causal role; nor could we say what
the action was, apart from specifying the intention to at least some
degree. An intention, unlike a cause, does not stand in an external,
contingent relation to an action. When Miss G. E. M. Anscombe
investigates the concept systematically in Intention, the whole
discussion moves away from any kind of explanation in terms of
causality, a topic to which Miss Anscombe alludes only in rare
passages. One, but only one, reason why this is so can be brought
out by considering how either in the kind of case which Miss
Anscombe would classify as one of "mental causality" (I am
startled by a noise and jump) or in the hard case where I make a
decision and later act on it (with which Miss Anscombe does not
deal) we should be missing the point if we looked for a Humean
generalization to connect the noise and the jump or the decision
and the action. I can know that I jumped because of the noise or
1 Philosophical Investigations, Part I, para. 644.
The antecedents of action 201
that I acted because of the decision and know perfectly well that
the generalizations "Whenever there is a noise of that sort, I
jump in this way" and "Whenever I decide to do something, I
do it" are false. Hence this kind of explanation of these actions at
least must be in terms other than those of causality. Beginning
from this point, the argument is sometimes generalized in the
following way.
It is bodily movements which are to be causally explained and
not human actions. Human actions are made intelligible by re-
ference to intentions, purposes, decisions, and desires. These do
not function as causes. They do not function as causes for at least
two distinct reasons. The first is that to say "He did it because he
intended so-and-so," or "He did it because he decided to" or
"He did it because he wanted to" is in each case not necessarily
to refer to two separately identifiable events, the doing on the one
hand and the intending or desiring or deciding on the other.
There may be cases where we first frame an intention, come to a
decision, or experience a desire and then act; but the concepts of
intention, decision, and desire are equally applicable where the
action is itself the expression of intention, decision, or desire and to
refer to our intention, decision, or desire in either explanation or
justification of our action is not to refer to an antecedent event.
But a cause must, so it is argued, always be a separate event from
that which is its effect. So intentions, decisions, and desires can-
not be causes. Secondly, intentions, decisions, and desires cannot
be causes, for they are not causally but logically related to the
relevant actions. How do I know that this intention relates to this
action? Not by any observed correlation such as would be rele-
vant in the case of causality. But because both intention and
action are mine and the intention contained a description of the as
yet unrealized action. The action is related to the intention as
being what is described in the forming of the intention. "I'll have
another cigarette in ten minutes' time." When I light up in ten
minutes, I am faithful to my intention, but my intention has not
made me light up.
Considerations such as these are invoked to support one of two
theses. Either the weaker assertion is made which I have already
described, that actions are to be explained in terms of intentions
and kindred concepts and therefore, insofar as this is the case,
they are not to be explained causally ; or else the stronger assertion
that causal explanations are out of court altogether so far as actions
are concerned: Unfortunately, the most extended statement of this
case is ambiguous to a certain extent. Melden says:
202 Against the Self-Images of the Age
What I shall be concerned to deny ... is that the term "cause"
when employed in these sciences (physics and physiology) is
applicable to those matters which, familiarly and on a common-
sense level, we cite in order to explain action: the motives,
desires, choices, decisions, etc., of human beings. I do not,
of course, deny that there are appropriate senses of "cause"
which can be intelligibly employed in these cases.1
And again he writes:
Indeed, it must appear problematic at best that the physiological
psychologist who purports to be attempting to explain human
action is addressing himself to his ostensible subject matter. 2
While still later he says :
Here (in cases where I am predicting what someone whom I
know will do) nothing is hidden; it is because I understand him,
not because I am aware of events transpiring in some alleged
mechanism of his mind or body, that I am able to say what he
will do. 3
These quotations can lend themselves to an interpretation in
which all that Melden claims is that to explain actions citing
purposes, intentions, desires, and the like is not to assign causes
(in the Humean sense) or to another interpretation in which
Melden is claiming that causal interpretations of human actions
are ruled out of court altogether on conceptual grounds. The
second quotation-apart from the fence-sitting use of "it must
appear problematic at best"-seems to ensure that the latter is
meant, but what I will presently try to do is to show that while the
latter thesis is certainly untenable, even in the former interpreta-
tion, Melden's thesis needs amendment. Before that, however, an
even more radical version of this view must be considered. It is
more radical because it is more systematic. It arrives at the same
conclusions as Melden's, but it derives them from independent
foundations. The best-known exposition of this point of view is
Dr. Friedrich Waismann's in Language Strata. Waismann wished
to campaign against the view that language is unitary, all of a piece,
that truth, rationality, and meaningfulness are one and the same
for every sort of statement. Against this he urged the notion of
language as composed of different strata, each with its own criteria
1 Free Action, pp. 16-17.
1 Ibid., p. 72.
3 Ibid., p. 208.
The antecedentsof action 203
of truth and meaning. An expression may be ambiguous in that it
can figure in different contexts in different strata, and so take on
different meanings. With this thesis so far my present argument
does not require me to raise any questions. But Waismann then,
although he allows that there are relationships between strata,
characterizes the ambiguity of the word "action" in such a way as
to exclude relationship between the stratum in which it is proper
to speak of causes and that in which it is proper to speak of motives.
It follows that anything which can be explained by reference to
motives cannot be explained by reference to causes and vice versa.
In like manner we say that each stratum has a logic of its own
and that this logic determines the meaning of certain basic terms.
In some respects this is obvious. Whether a melody is a sequence
of air-vibrations, of a succession of musical notes, or a message
of the composer, depends entirely on the way you describe it.
Similarly, you may look at a game of chess, or on the pattern of a
carpet from very different aspects and you will then see in them
very different things. Notice how all these words-"melody,"
"game of chess," etc.-take on a systematic ambiguity according
to the language stratum in which you talk. The same applies to
"doing a sum," "writing a letter," or to any action indeed. An
action may be viewed as a series of movements caused by some
physiological stimulus in the "Only rats, no men" sense; or as
something that has a purpose or meaning irrespective of the way
its single links are produced. An action in the first sense is deter-
mined by causes, an action in the second sense by motives or
reasons.It is generally believed that an action is determined both
by causes and by motives. But if the causes determine the action,
no room is left for motives, and if the motives determine the
action, no room is left for causes. Either the system of causes is
complete, then it is not possible to squeeze in a motive; or the
system of motives is complete, then it is not possible to squeeze
in a cause. "Well now, do you believe that if you are writing a
letter you are engaged in two different activities?" No; I mean
that there are two different ways of looking at the thing; just as
there are two different ways of looking at a sentence: as a s~ries
of noises. produced by a human agent; or as a vehicle for thought.
For a series of noises there may be causes but no reasons; for a
series of words expressing thought there may be reasons, but no
causes. What is understood is that the word "action" has a
systematic ambiguity. 1 (pp. 30-31).
1 Language Strata, pp, 30--31.
Against the Self-Images of the Age
V
What is at stake in these arguments? Not only philosophical
clarity, but also the question of the nature of the human sciences.
For if philosophical argument can show that actions cannot have
causes, then a good deal of science is fatally confused, since
scientists do in fact attempt to offer causal explanations of action.
Some physiologists have, indeed, done us a disservice by offering
explanations of reflexes and calling these explanations of action,
but in more than one field there appear to be genuine, if tentative,
causal explanations of action. I refer to criminology and also to
the study of the effects of drugs. (The study of hypnosis is
interesting, but raises special issues.)
Some changes in the chemistry of the body which are brought
about by taking drugs correlate with highly specific alterations in
behavior. More than this, we can alter the way in which people
behave by inducing such changes in body chemistry. These
changes range from the medical use of insulin in highly artificial
laboratory experiments to buying a man a drink. What is correlated
with the chemical change is a type of action and not just a type of
bodily movement. That is to say, the framing of intentions,
deliberations, reflection on wishes and desires and the like all play
their normal roles. It may be that in many cases the type of action
which is produced by the chemical changes cannot be narrowly
specified. That is, we can say that to give this type of man this
type of drug will make him act more irritably or unscrupulously
or excitably rather than specify in more detail what he will do.
But in these cases we are none the less involved in explaining
behavior.
Again, in criminology the work that has shown that there is a
hereditary element in criminality is much to the point. The key
studies on inherited characteristics in human beings are those
which compare the degree to which such characteristics are shared
by two siblings in the case of monozygotic and dizygotic twins
respectively, for it would seem an unassailable conclusion that a
clearly higher concordance in cases of monozygotic twins would
indicate a hereditary factor. This is how the existence of a
hereditary factor in tuberculosis was established for a study in
New York in 1934, for example, which showed a 63 per cent
concordance in monozygotic twins and only an 18 per cent con-
cordance in dizygotic, while one in London in 1957 showed a 30
The antecedentsof action 205
VI
Let us begin with the most general form of the argument that
actions cannot have causes, Waismann's. It is beyond the scope of
this essay to question in general terms the widely influential, but
profoundly misleading view of language contained in Waismann's
paper. What one must note is that unless there were expressions
and criteria which transcended the divisions between his language
strata, he could not distinguish them in the way he does. He can,
for instance, recognize, and has to recognize in order to specify
the ambiguity of "action" in the way he does, that certain move-
ments caused by physiological stimuli use the movements which
are the "single links" of this particular action and of no other.
So that we are able to say "These movements" (one stratum of
language) "belong to this action" (a quite different stratum). It
follows that statements are not necessarily confined to one particular
logical order or type or stratum (however these are specified).
And now we have to note that an expression that cannot be so
confined is the word "cause" and its logical kith and kin. For
Waismann speaks of the bodily movement by which an action is
206 Against the Self-Images of the Age
"produced." And "produce" is certainly a causal verb. Moreover,
he speaks of a melody as though to speak of a sequence of air
vibrations is to remain within one language stratum, but to speak
of a succession of musical notes is to move to another. But clearly
one cause, in a perfectly acceptable and unambiguous sense of
"cause," of a set of notes succeeding one another in a piece of music
to which we are listening is precisely the sequence of air vibrations
which the orchestra have produced. Without having read Wais-
mann, we might well want to say that the notion of color is of a
different logical order from that of a wavelength of light. But we
unhesitatingly explain alterations in color as caused by changes in
the wavelength of light. So that, although the notion of bodily
movements may be of a different logical order from that of an
action, it certainly cannot follow that the word "cause" is restricted
to the stratum to which "bodily movement" belongs and denied to
the stratum to which "action" belongs.
If we then disallow Waismann's contentions, the arguments that
actions cannot have causes are best dealt with not by denying the
importance of the type of example to which such arguments
appeal, but by considering counter-examples. The suggestion will
be that all the ordinary senses of causality apply in some cases to
human actions and that therefore the "ordinary language" use of
"cause" in this connection is by no means as remote from Humean
causality as some suggest. (For an "ordinary language" use, con-
sider: "The Minister, receiving the Woman at her father's or
friend's hands, shall cause the Man with his right hand to take the
Woman by her right hand ... " [Book of Common Prayer, Form
of Solemnization of Matrimony.]
The first example is of giving a reason or affording a motive as
causes. I may discover that when you are in a certain frame of
mind I can get you to act by giving you information which
affords a motive or a reason. Your action bears testimony to the
fact that it was this motive or reason on which you were acting
(as returning a ring with a reproachful letter is testimony that the
girl's motive arises from her information about the man's behavior).
Thus the connection between affording the motive and the
action is not one of a Humean kind; we do not depend on a
universal generalization of whose truth we need to be assured in
order to make the connection. Even if another occasion affording
the same kind of motive does not produce the same action, we
should not have grounds for doubting what caused the girl to act
in the way she did on the first occasion. And the word "caused" is in
place precisely because of our third sense of causality. Affording a
The antecedents of action
motive or a reason is performing a separately identifiable_and
desirable act, the performance of which is a lever that produces as
its effects an action. And it is quite compatible with the thesis that
motives, reasons, decisions, and intentions cannot be causes that
affording motives, giving reasons, giving grounds for decisions,
and for framing intentions can be. Nor is this merely something
that others can do to me; I can in deliberation do this to myself.
(This is not to be committed to the view that deliberation is
always conversation with oneself, but only to the view that it can
on occasion be.)
Secondly, "the" cause of an action may, like "the" cause of an
accident, be a necessary condition, the satisfaction of which is with
other circumstances sufficient to produce the action. An insult
may not make me violent when I am sober; and when I am even
mildly drunk I may be extremely pleasant except when and until I
am insulted. So the insult plays in relation to the action the part
that the icy patch on the road plays in relation to the accident.
Thirdly, we are already well on the way to formulating explana-
tions of actions in terms of Humean causality. I am puzzled by
why I become angry when playing cards. Both others and myself
presently observe that it occurs five minutes after I have started to
lose. This connection is uniformly observed to hold until I
become aware of it. People who wish to make me angry have
learned that I become angry and perform angry actions if they
bring about my defeat at cards. This is a perfect case of Humean
causality and nothing is affected if I change my behavior on
discovering its cause. For the generalization which needed to be
discovered by observation was that "Whenever I am losing at cards,
and so long as I do not know what is going to happen to my
behavior as a result, I shortly after become angry." Or it may
be that I cannot alter my behavior. Obviously, it does not follow
that I am inevitably going to be angry; but if I wish to avoid
angry behavior then I must not lose at cards, and probably I
must not play cards.
These examples only skim the topic of the causality of action.
What they do show is the danger of any generalization of the
form "Actions cannot have causes" or even "Actions cannot have
Humean causes." Such generalizations are necessarily as erroneous
as were the generalizations of the eighteenth-century mechanists
who thought of every action as caused. What we need is a much
fuller characterization of the concept of the human person in which
the role of both causes on the one hand and of motives, reasons,
and intentions on the other will become clear. But about two
208 Against the Self-Images of the Age
distorting features of the discussion hitherto we can now perhaps
become clear.
The first is that the dichotomy "logical connection" or "causal
connection" is much too easy, here as elsewhere. Consider the
kind of case where an insult always leads to taking offense. "Every
Celt when insulted uses whatever weapons lie to hand" can be the
expression of a good Humean generalization (even if false). Now
the description of the first action as "an insult" and of the second
as "taking offense" brings them under descriptions which are
conceptually and internally related. But the two events are
separately identified and we can correlate them. We know what it
would be for the causal generalization to be falsified by an insult not
causing offense or offense being taken without insult. The root
error here is to think of actions as standing in relationship to the
agents' motives and reasons or to other agents' behavior inde-
pendently of the alternative forms of description under which
behavior can fall.
The second distorting factor is the fear of determinism. This
perhaps springs from accepting a determinist view of the Hume-
Mill concept of causality. But to show that an action is caused is
not necessarily to show that it must have happened, that the
agent could not alter what he did. For to assign a cause to a
happening is to go some way to informing us both how to produce
and how to inhibit the happening in question. It follows from
this that to assign causal explanations to actions is not to show
that the actions in question are inevitable and unalterable. Nor
does it even follow that if the explanations in question are explana-
tions of my actions, I cannot alter them. But it certainly does
follow that the more I know about possible and actual causal
explanations of my behavior the more likely I am to be able to
intervene successfully and control what I do. Free, responsible,
controlled behavior is, then, behavior where I have at least the
possibility of successful intervention (though this is to state only a
necessary, and not a sufficient, condition for being entitled to
characterize a piece of behavior in this way).
This argument needs one addition. My freedom as an agent
depends upon my ability to frame intentions which are capable of
being implemented. This capability is dependent on the reliability
of my beliefs about the world and about myself: it is not just that
given motives, desires, and intentions of a certain sort, I act. A
presupposition of successful action is a knowledge of what I will
do unless I intervene in various ways. So the concept of intention
cannot be understood in isolation from the role of belief and
The antecedents of action 209
knowledge in our behavior. The way in which this is so is brought
out most clearly by the argument of Stuart Hampshire in Thought
and Action, especially Chapter 3. Hampshire uses the distinction
between intention and prediction in a much more illuminating way
than it is used by those who are trying to separate action and
causation. For, as I argued earlier, there is a positive connection
between intention and prediction. Unless I am able to predict what
will happen if I do and again what will happen if I do not frame a
given intention, I am in no position to frame intentions at all.
Thus, what I can intend depends upon what I can predict, and
the dependence of the concept of human action upon the concept of
intention does not exclude the possibility of prediction based on
causal explanation from the realm of human action, but actually
depends upon that possibility. That others can predict my actions
does not matter unless they are able to predict what these are, no
matter what my intentions are. My freedom consists, as Hampshire
has argued, not in my unpredictability but in my ability to form
clear intentions and to implement them. And this freedom depends
on my ability to intervene in causal sequences, including those
which have resulted in parts of my own behavior to date.
The mistake that we might make in conclusion would be to
suppose that because my main argument has been an attack on
the generalizations of others, nothing definitive follows from it.
I followed through the attack on the doctrine of acts of will and
showed that the corollary to its destruction was the need to
elaborate a much more complex view of the person. But some
very simple conceptual truths still need emphasis, and one of them
at least belonged to the view of the will which Ryle and Melden
attempted to destroy.
The exponents of the traditional doctrine of acts of will were
clearly wrong to hold that every act of a rational agent is preceded
by an act of will which is its cause. They took what they thought
to be the paradigm case of rational action, deliberation leading
to conscious decision which issues in action, and supposed
that the characteristics of the paradigm case must hold in every
case. But were they wrong in their characterization of the
paradigm? Where I act without deliberation or on impulse,
where I provide one of the cases which appear to Ryle and Melden
to destroy the traditional doctrine, what makes me responsible
for what I do? Or where causal explanations of my anger are in
place, what, if anything, makes me responsible? Presumably, that I
could, had I reason so to frame an intention, decide not to do what
I in fact do, not to let my impulses have their way or not to be
210 Against the Self-Images of the Age
angry; that I could deliberate (the "could" means here that it
makes sense to speak of my deliberating; in many actual occasions I
might have no time or opportunity to deliberate) and decide on
some other course of action. In other words, that I could perform
an act of will in the traditional sense. If we read Hobbes, Hume
and Kant as characterizing not the necessary prerequisites for
something to be classed as an action but as characterizing the type
of action which one must be able to perform on occasion, the type
of intervention, inhibiting one course of action or unleashing
another, which one must be able to make on occasion if one is to
be classed as a responsible and rational agent at all, then the
arguments of Ryle and Melden become irrelevant. For acts of will
are, as the traditional authors clearly state, the familiar and un-
assailable processes of resolving, deciding, and intending and not
the mysterious and occult "volitions" with which Ryle and Melden
make so much play. There is nothing here with which novelists
and ordinary agents are not familiar in their everyday transactions ;
and there is no requirement that every action shall be preceded by
a volition, which may result in an infinite regress. There is only
the requirement that we shall recognize that it is in virtue of what
they can be and not of what they always are that men are called
rational animals.
The idea of a social science 1
I
"Wittgenstein says somewhere that when one gets into philo-
sophical difficulties over the use of some of the concepts of
our language, we are like savages confronted with something
from an alien culture. I am simply indicating a corollary of this:
that sociologists who misinterpret an alien culture are like philo-
sophers getting into difficulty over the use of their own concepts."
1 © Aristotelian Society, 1967. Reprinted from the Aristotelian Soeiety
Supplementary Volume 1967, pp. 95-u4.
2 The Idea of a Social Science (London: Routledge and Kegan Paul,
1958).
3 See, for example, Richard Rudner, The Philosophy of Social Science
II
"A regularity or uniformity is the constant recurrence of the
same kind of event on the same kind of occasion; hence statements
of uniformities presuppose judgments of identity. But ...
criteria of identity are necessarily relative to some rule: with the
corollary that two events which count as qualitatively similar from
the point of view of one rule would count as different from the
point of view of another. So to investigate the type of regularity
studied in a given inquiry is to examine the nature of the rule
according to which judgments of identity are made in that inquiry.
Such judgments are intelligible only relatively to a given mode
of human behavior, governed by its own rules" (pp. 83-84).
This passage is the starting point for Winch's argument that
J. S. Mill was mistaken in supposing that to understand a social
institution is to formulate empirical generalizations about regu-
larities in human behavior, generalizations which are causal and
explanatory in precisely the same sense that generalizations in
the natural sciences are. For the natural scientist makes the
relevant judgments of identity according to his rules (that is,
the rules incorporated in the practice of his science) ; whereas the
social scientist must make his judgments of identity in accordance
with the rules governing the behavior of those whom he studies.
Their rules, not his, define the object of his study. "So it is quite
mistaken in principle to compare the activity of a student of a
form of social behavior with that of, say, an engineer studying
the working of a machine. If we are going to compare the social
214 Against the Self-Images of the Age
student to an engineer, we shall do better to compare him to an
apprentice engineer who is studying what engineering-that is,
the activity of engineering-is all about" (p. 88).
What the type of understanding which Winch is commending
consists in is made clearer in two other passages. He says that
although prediction is possible in the social sciences, it "is quite
different from predictions in the natural sciences, where a falsified
prediction always implies some sort of mistake on the part of
the predictor: false or inadequate data, faulty calculation, or
defective theory" (pp. 91-92). This is because "since understand-
ing something involves understanding its contradictory, someone
who, with understanding, performs X must be capable of envis-
aging the possibility of doing not-X" (p. 91 ). Where someone
is following a rule, we cannot predict how he will interpret what is
involved in following that rule in radically new circumstances;
where decisions have to be made, the outcome "cannot be definitely
predicted," for otherwise "we should not call them decisions."
These points about prediction, if correct, reinforce Winch's
arguments about the difference between the natural sciences and
the social sciences. For they amount to a denial of that symmetry
between explanation and prediction which holds in the natural
sciences. (It has been argued often enough that this symmetry
does not hold in the natural sciences; Professor Adolf Griinbaum's
arguments in Chapter 9 of the Philosophy of Space and Time
seem a more than adequate rebuttal of these positions.) But when
we consider what Winch says here about decision, it is useful to
take into account at the same time what he says about motives
and reasons. Winch treats these as similar in this respect: that
they are made intelligible by reference to the rules governing
the form of social life in which the agent participates. So Winch
points out that "one can act 'from considerations' only where
there are accepted standards of what is appropriate to appeal to"
(p. 82) and argues against Ryle that the "law-like proposition"
in terms of which someone's reasons must be understood concerns
not the agent's disposition "but the accepted standards of reason-
able behavior current in his society" (p. 81).
From all this one can set out Winch's view of understanding
and explanations in the social sciences in terms of a two-stage
model. An action is first made intelligible as the outcome of
motives, reasons, and decisions; and is then made further in-
telligible by those motives, reasons, and decisions being set in the
context of the rules of a given form of social life. These rules
logically determine the range of reasons and motives open to a
The idea of a social science 215
given set of agents and hence also the range of decisions open to
them. Thus Winch's contrast between explanation in terms of
causal generalizations and explanations in terms of rules turns
out to rest upon a version of the contrast between explanations
in terms of causes and explanations in terms of reasons. This latter
contrast must therefore be explored, and the most useful way
of doing this will be to understand better what it is to act for a
reason.
Many analyses of what it is to act for a reason have written
into them an incompatibility between acting for a reason and
behaving from a cause, just because they begin from the apparently
simple and uncomplicated case where the action is actually
performed, where the agent had one and only one reason for
performing it, and where no doubt could arise for the agent as to
why he had done what he had done. By concentrating attention
upon this type of example, a basis is laid for making central to the
analyses a contrast between the agent's knowledge of his own
reasons for acting and his and others' knowledge of causes of his
behavior. For clearly in such a case the agent's claim that he did
X for reason Y does not seem to stand in need of any warrant
from a generalization founded upon observation; while equally
clearly any claim that one particular event or state of affairs
was the cause of another does stand in need of such a warrant.
But this may be misleading. Consider two somewhat more com-
plex cases than that outlined above. The first is that of a man
who has several quite different reasons for performing a given
action. He performs the action; how can he as agent know whether
it was the conjoining of all the different reasons that was sufficient
for him to perform the action or whether just one of the reasons
was by itself alone sufficient or whether the action was over-
determined in the sense that there were two or more reasons, each
of which would by itself alone have been sufficient? The problem
arises partly because to know that one or other of these possibilities
was indeed the case entails knowing the truth of certain unfulfilled
conditionals.
A second case worth considering is that of two agents, each
with the same reasons for performing a given action; one does
not in fact perform it, the other does. Neither agent had what
seemed to him a good reason or indeed had any reason for not
performing the action in question. Here we can ask what made
these reasons or some subset of them productive of action in the
one case, but not in the other? In both these types of case we
need to distinguish between the agent's having a reason for
8
216 Against the Self-Images of the Age
performing an action (not just in the sense of there being a reason
for him to perform the action, but in the stronger sense of his
being aware that he has such a reason) and the agent's being
actually moved to action by his having such a reason. The im-
portance of this point can be brought out by reconsidering a
very familiar example, that of post-hypnotic suggestion.
Under the influence of post-hypnotic suggestion a subject
will not only perform the action required by the hypnotist, but
will offer apparently good reasons for performing it, while quite
unaware of the true cause of the performance. So someone
enjoined to walk out of the room might, on being asked why he
was doing this, reply with all sincerity that he had felt in need of
fresh air or decided to catch a train. In this type of case we would
certainly not accept the agent's testimony as to the connection
between reason and action, unless we are convinced of the un-
truth of the counter-factual. "He would have walked out of the
room, if no reason for doing so had occurred to him" and the
truth of the counter-factual, "He would not have walked out
of the room, if he had not possessed some such reason for so
doing." The question of the truth or otherwise of the first of
these is a matter of the experimentally established facts about
post-hypnotic suggestion, and these facts are certainly expressed
as causal generalizations. To establish the truth of the relevant
generalization would entail establishing the untruth of the second
counter-factual. But since to establish the truth of such causal
generalizations entails consequences concerning the truth or
untruth of generalizations about reasons, the question inevitably
arises as to whether the possession of a given reason may not be the
cause of an action in precisely the same sense in which hypnotic
suggestion may be the cause of an action. The chief objection
to this view has been that the relation of reason to action is internal
and conceptual, not external and contingent, and cannot therefore
be a causal relationship ; but although nothing could count
as a reason unless it stood in an internal relationship to an action,
the agent's possessing a reason may be a state of affairs identifiable
independently of the event which is the agent's performance
of the action. Thus it does seem as if the possession of a reason by
an agent is an item of a suitable type to figure as a cause, or an
effect. But if this is so then to ask whether it was the agent's
reason that roused him to act is to ask a causal question, the true
answer to which depends upon what causal generalizations we
have been able to establish. This puts in a different light the
question of the agent's authority as to what roused him to act;
The idea of a socialscience 217
for it follows from what has been said that this authority is at
best prima facie. Far more of course needs to be said on this and
related topics; but perhaps the argument so far entitles us to
treat with skepticism Winch's claim that understanding in terms
of rule-following and causal explanations have mutually exclusive
subject matters.
This has obvious implications for social science, and I wish
to suggest some of these in order to provide direction for the rest
of my argument. Clearly if the citing of reasons by an agent,
with the concomitant appeal to rules, is not necessarily the citing
of those reasons which are causally effective, a distinction may be
made between those rules which agents in a given society sincerely
profess to follow and to which their actions may in fact conform,
but which do not in fact direct their actions, and those rules
which, whether they profess to follow them or not, do in fact
guide their acts by providing them with reasons and motives for
acting in one way rather than another. The making of this
distinction is essential to the notions of ideology and of / alse
consciousness,notions which are extremely important to some
non-Marxist as well as to Marxist social scientists.
But to allow that these notions could have application is to
find oneself at odds with Winch's argument at yet another point.
For it seems quite clear that the concept of ideology can find
application in a society where the concept is not available to the
members of the society, and furthermore that the application of
this concept implies that criteria beyond those available in the
society may be invoked to judge its rationality; and as such it
would fall under Winch's ban as a concept unsuitable for social
science. Hence there is a connection between Winch's view that
social science is not appropriately concerned with causal generali-
zations and his view that only the concepts possessed by the mem-
bers of a given society (or concepts logically tied to those concepts
in some way) are to be used in the study of that society. Further-
more, it is important to note that Winch's views on those matters
necessarily make his account of rules and their place in social
behavior defective.
III
The examples which Winch gives of rule-following behavior
are very multifarious: games, political thinking, musical com-
position, the monastic way of life, an anarchist's way of life, are all
218 Against the Self-Images of the Age
cited. His only example of non-rule-governed behavior is "the
pointless behavior of a berserk lunatic" (p. 53), and he asserts
roundly "that all behavior which is meaningful (therefore all
specifically human behavior) is ipso facto rule-governed." Winch
allows for different kinds of rules (p. 52); what he does not
consider is whether the concept of a rule is perhaps being used so
widely that quite different senses of rule-governed are being con-
fused, let alone whether his account of meaningful behavior can
be plausibly applied to some actions at all.
If I go for a walk, or smoke a cigarette, are my actions rule-
governed in the sense in which my actions in playing chess are
rule-governed? Winch says that "the test of whether a man's
actions are the application of a rule is . . . whether it makes
sense to distinguish between a right and a wrong way of doing
things in connection with what he does." What is the wrong
way of going for a walk? And, if there is no wrong way, is my
action in any sense rule-governed? To ask these questions is to
begin to bring out the difference between those activities which
form part of a coherent mode of behavior and those which do not.
It is to begin to see that although many actions must be rule-
governed in the sense that the concept of some particular kinds
of action may involve reference to a rule, the concept of an action
as such does not involve such a reference. But even if we restrict
our attention to activities which form part of some coherent
larger whole, it is clear that rules may govern activity in quite
different ways. This is easily seen if we consider the variety of
uses to which social scientists have put the concept of a role and
role concepts.
Role concepts are at first sight peculiarly well-fitted to find a
place in the type of analysis of which Winch would approve.
S. F. Nadel wrote that "the role concept is not an invention of
anthropologists or sociologists but is employed by the very
people they study," and added that "it is the existence of names
describing classes of people which make us think of roles." It
would therefore be significant for Winch's thesis if it were the
case that role concepts had to be understood in relation to causes,
if they were to discharge their analytic and explanatory function.
Consider first a use of the notion of role where causal questions
do not arise. In a society such as ours there are a variety of roles
which an individual may assume or not as he wills. Some
occupational roles provide examples. To live out such a role
is to make one's behavior conform to certain norms. To speak
of one's behavior being governed by the norms is to use a sense
The idea of a socialscience 219
IV
The positive value of Winch's book is partly as a corrective to
1 Comparative Functionalism (Berkeley: University of California Press,
I
A man who yesterday chatted pleasantly with an academic
colleague today crosses the street to avoid meeting him. Why?
He has in the interval read a review of his latest book by this
colleague and resents what he takes to be his unjust verdict. He
expresses his resentment by crossing the road. Suppose that he
knew that this colleague had a peculiar love of a rare fruit which
he could only procure at one store; he might then buy up the entire
stock of that fruit and so express his resentment. Or suppose that
he knows that what his colleague prizes is an invitation to a
particular party; he might then express his resentment by inter-
cepting the invitation. Crossing a road, buying up fruit, stealing
mail: these actions have. nothing in common with each other and
yet they can all express resentment. Precisely because there is no
characteristic which they have to possess in order to function as
expressions of resentment, precisely because, except as possible
expressions of resentment, there is no reason for including these
items in the list rather than any others, it seems plausible to suppose
that any action whatsoever can function as an expression of
resentment.
Consider now three possible criticisms of this argument. Surely,
it might be argued, all such items of behavior do fall under some
one single description other than "resentful" ; they are all, for
Against the Self-Images of the Age
example, items of hostile behavior. This is of course true, but it
does not affect my point. First of all, just as any action at all can
be an expression of resentment, any action at all can be a hostile
action. Crossing a road, buying up fruit, stealing mail can all be
hostile in specific contexts, just as they can be resentful in specific
contexts. To this it might be objected that there are some actions
which could not be hostile, if by that is meant "directed to the
harm of others" ; how could suicide be in this sense hostile? The
answer is, of course, that there is a well-recognized class of suicides,
the so-called "revenge" suicides ("I'll kill myself and that will
teach them a lesson!") where the point of the suicide is precisely
that it is a hostile action. Moreover, the variety of types of action
which can be characterized as expressions of resentment can also
be characterized as hostile just because "hostile" is partially
synonymous with "resentful," in turn because resentment is a
species of hostility. So that the fact that we can find another
description under which all these actions fall does not in any way
show that we have found an additional common characteristic
shared by all these action~.
A second objection to my thesis might be that I am able to
understand all these actions as expressions of resentment only by
establishing a context of a given kind. It is not crossing the road
or buying fruit or stealing mail that is the expression of resentment,
but crossing-the-road-to-avoid-speaking-to-someone-who-has-un-
justly-criticized-me or buying-fruit-especially-desired-by-some-
one-who-has-unjustly-criticized-me, and so on. This is in a way
correct. We are only able to construe the actions in question as
behavior expressive of resentment by connecting them with the
reasons that the agent has for doing what he does. But these reasons
cannot be identified with what the agent does nor are they ex-
hibited in doing what he does. Certainly it is only because the
action is done for a certain kind of reason that it is an expression of
resentment. But to say this is to say that qua action, and qua the
action that it is, the relation between the action and the resentment
is purely contingent. To make the point in this way does perhaps
help us to understand better why many philosophers may have
thought otherwise. They may have envisaged the action under
some description, such as the descriptions above, that links the
action to the reason for performing it and passed from asserting
truly that such a description is conceptually connected with the
characterization of the emotion in question as resentment to
asserting falsely that the action itself was not merely contingently
connected to the resentment.
Emotion, behaviorand belief 233
A third point might be raised not so much as an objection to my
initial thesis as an objection to drawing possibly illegitimate con-
clusions from it. For it might be suggested that while what I have
said is true of the emotion of resentment, it is not true of emotions
in general. What is special about the emotion of resentment? A
man cannot be said to resent something unless he has a particular
type of belief. He must believe that he has been wronged in the
light of what it is established that a man in his position has a right
to expect. Unless he has such a belief, what he feels may be
characterized perhaps as anger but not as resentment. What then
of anger? Is there some belief which a man who is said to be angry
must possess? It is clearly the case that usually and characteristi-
cally a man who is angry believes that something has been done
that is harmful to or an affront to himself or his interests or those
about whom he cares. But is it not sometimes the case that a man
just feels angry? And, therefore, does not anger perhaps differ in a
not irrelevant way from resentment?
The relevance of this point to the argument is as follows. In the
case of resentment it is because the circumstances, tastes, and
other relevant characteristics of the person against whom resent-
ment is directed are indefinitely variable that the actions which
may express that resentment are indefinitely variable. Hence, the
indefinite variability of the actions which express resentment is
connected with the belief which a man to whom resentment is
correctly ascribed must possess. But if in the case of anger there
is no such belief, then may not anger be connected with behavior
expressive of anger in some way quite different to that in which
resentment is connected with behavior expressive of resentment?
It is important to stress first that where anger is divorced from
the belief that usually and characteristically accompanies it,
(namely, the belief that some identifiable person has done some
identifiable harm), we are confronted not with anger as it basically
is (that is, with the emotion in a pure form, so to speak), but
rather with an uncharacteristic and marginal case, which is less
easily intelligible to us than anger in its usual form and which we
understand by its resemblance to these forms. (Try to imagine a
culture in which everyone is all the time in a rage with no one in
particular about nothing in particular, but is never angry with
specific individuals about specific harms. I am inclined to think that
we would treat this as a different emotion.)
Secondly, anger in these special cases is not in fact unaccom-
panied by belief; it is just that the belief is expressed in sentences
containing more variables. The belief that someone or anyone has
234 Against the Self-Images of the Age
done something harmful or affronting to me, although I know not
who or what, is still a belief, and the belief that connects the
feelings accompanying it to other feelings of anger. Hence anger,
like resentment, is connected with a belief and although the belief
is a less complex one, the persons or actions against whom anger
is directed are as indefinitely variable as are the objects of resent-
ment. Hence also the forms of behavior by means of which anger
too may be expressed are indefinitely variable.
I take it therefore that the lack of any necessary connection
between emotion and behavior holds in the case of anger as well
as in that of resentment. But if anger and resentment resemble
each other in this way, how do they differ? It seems plausible to
suggest that the only difference is in the beliefs of the agents in
question. The emotion of anger involves the belief that someone,
whether I deserve it or not, has done harm or has affronted me or
my interests or those about whom I care; the emotion of resent-
ment involves the belief that someone has done undeserved harm
or offered an undeserved affront to me or my interests. The dif-
ference is not in the phenomenological feel of the two emotions:
introspective reports do not reveal different sensations in the case
of anger from those present in the case of resentment. Nor is the
difference in the behavior through which each is expressed, since
each may be expressed in the same behavior. The beliefs alone
provide a difference.
To this it may be retorted that the relationship of anger to
resentment is a special case. For after all resentment is a species of
anger. It may therefore be true that the felt quality of the emotion
and the behavior do not differ in this case. But if, instead of asking
for the difference between anger and resentment, we were to
inquire what is the difference between anger and elation or be-
tween resentment and gratitude, the same would not hold.
Consider resentment and gratitude. To feel grateful is to feel
pleased that someone has done more for your good or for the good
of those about whom you care than you had a right to expect.
But like resentment any kind of behavior may express gratitude;
the fact that one emotion is one of pleasure in something and the
other of displeasure at something does not entail that the very
same behavior may not express gratitude which expresses resent-
ment. If I am grateful to you for what you have done and I know
that you resent what someone else has done, I may express my
gratitude to you by doing to him what if you did it to him would
express your resentment. Hence the difference between gratitude
and resentment is not a difference in behavior. I take it that it is
Emotion, behaviorand belief 235
also not a difference in the felt quality of inner states. The feelings
of a man in the extremity of an emotion such as gratitude is
customarily described by novelists in ways that are remarkably
like the ways in which the feelings of a man in the extremity of an
emotion such as fear are described. His throat goes dry, his temples
throb, his pulse rate rises, his eyes prick with tears, and so on.
The physiological symptoms of emotion seem remarkably constant
and the physiologists seem to agree with the novelists about this.
If I conclude then that the difference between emotions lies in
the belief and not in the behavior, I cannot possibly identify
emotions with patterns of behavior. But a behaviorist might try
at this point to recover his position by an argument of more
general import. For he might contend that the notion of belief
itself is to be analysed in terms of behavior, asserting that "X
believes that p" is logically equivalent to or means the same as
"X has a disposition to manifest certain patterns of behavior" and
also that to manifest a belief is to manifest patterns of behavior, so
that the appropriate evidence which would warrant the assertion
that "X believes that p" is that X has on occasion exhibited the
relevant behavior. In order for these assertions to be defensible,
the behavior in question would have to include what X says as well
as what X does, what some behaviorists have quaintly called
"linguistic behavior." More specifically it is what X asserts and not
just what X says which has to be included in the list of relevant
items: unless on certain types of possible occasions a man was
prepared to assert that p and to deny that p he could not be said
to believe that p. It follows, therefore, that if the notion of belief
is to be analysed in terms of behavior, the behavior in terms of
which it is analysed must include acts of assertion and denial.
But not what is it to assert that p? It is to give one's hearers or
readers to understand that one believes that p and that p is worthy
of belief. That is to say, the notion of assertion has to be explained
by referring to the notion of belief. So it turns out that the notion
of belief has not been analysed in terms of behavior, for the type of
behavior to which reference has to be made in the course of the
analysis can itself only be understood by referring to the concept
of belief. The concept of belief is at least as fundamental a concept,
and possibly a more fundamental concept than that of behavior.
To this it might be answered that to assert and to deny are just
forms of behavior. There is clearly a sense in which this is true.
If I say of someone that he behaved disgracefully in denying that
he was to blame, I say something intelligible to every user of
standard English. But in this idiomatic sense of "behavior" we
Against the Self-Images of the Age
cannot give a behaviorist account even of behavior. For to assert
of someone that he asserts that p is to go beyond saying that he
utters the sentence "p" even if he did in fact assert that p by
uttering the sentence "p." It is to construe his utterance in terms
of his intention in uttering the sentence. The intentions that
inform that utterance, like all intentions, presuppose beliefs. So
that we have once again, in the course of trying to analyse the
concepts of belief, been brought back to it.
Finally it ought to be noticed that in discussing behaviorism
I have not ascribed to the behaviorist that extreme view which
equates behavior with physical movement. When I argued that
there are at least certain emotions which can be expressed by any
behavior whatsoever, my examples-those of crossing the road,
buying fruit, and stealing mail-were all examples of actions and
their descriptions were descriptions which specified an intention
embodied in what was done. If the behaviorist wants to insist
that he means more than this by behavior, that in his view be-
havior has not been characterized adequately until it is charac-
terized by the emotions it expresses, if any, and that therefore
resentful behavior for example is just a species of behavior and
not behavior informed by something other than that behavior,
namely an emotion, his thesis becomes trivial. If he wants, on the
other hand, to insist that he means less than this by "behavior,"
perhaps equating behavior with physical movement, then my
preceding arguments hold with as great or greater force than they
do against behavior equated with human action.
The outcome of my arguments is then that behavior stands in
indirect and complex relationship to emotions. Without accepting
classical introspectionism, the introspectionist and indeed ordinary
language usage according to which we speak of emotions as lying
behind behavior and as being concealed as much as revealed by it
does not seem exaggerated. But if this is so, what are we to say
about our knowledge of the emotions of others? How far can we
know what they feel?
II
Paul Ziff in his About Behaviorism considers the contention that
"You can in principle if not in fact always find out whether or not
I am behaving in certain ways. But you cannot even in principle
always find out whether or not I am angry," and even waiving all
difficulties about the locution "you can in principle find out" finds
Emotion, behaviorand belief 237
what he takes to be two fatal objections to it. One which I shall
not consider, since I agree with it, is that it is fal~e that you can in
principle always find out how I am behaving. The other objection
he puts as follows: "You can in principle always find out whether
or not I am angry because I can tell you. Hence you need attend
only to my verbal behavior. (I assume that it would be generally
odd to speak of my being mistaken about whether or not I am
angry.) To suppose that you cannot in principle find out whether
or not I am angry would be to suppose that I cannot in principle
tell you whether or not I am angry. I find such a supposition
unintelligible." Agreeing with finding this last supposition un-
intelligible, I still want to disagree with the main point.
From the fact that, if I am angry, I can always tell you that I am
angry, it does not follow that from what I say you can always tell
if I am angry or not. The reason for this is that I can always
deceive you by lying or otherwise misleading you and that I can
always simply refuse to reveal what I feel. The use of "can in
principle," which Ziff overgenerously adopts from the anti-
behaviorists whom he is criticizing, is one source of trouble. For it
may lead us to concede not only that I may always when angry
tell you that I am, but also that your knowledge of my anger is
unproblematic. But it isn't.
When Ziff talks of verbal behavior, he may be suggesting that
what I say about my emotions stands to my emotions in the same
relationship that the behavior which expresses my emotions
stands to my emotions. But this is surely false either in a behaviorist
or in a nonbehaviorist view. In a behaviorist view it is false because,
in that view, my having an emotion consists in my exhibiting a
certain pattern of behavior: the behavior which expresses the
emotion is the emotion. But when I tell you what I feel, I do not
express the emotion, I report it. Or rather, I may be expressing the
emotion in the act of reporting it or not, but I am certainly report-
ing it. This is partly a matter of to whom I am speaking. If I am
angry with you and I say "I am angry with you," I am undoubtedly
expressing my anger in the act of reporting it. But if I am angry
with you and say to someone else "I am angry with Smith," then
it would be odd to say that I was expressing my anger. It is in any
case utterances and neither sentences nor statements that are
expressive of emotion, and utterances may certainly stand to an
emotion just as other behavior stands to it. But what I say when I so
utter does not stand in the same relationship to the emotion, and as
a counterpart of my thesis that in certain cases at least emotions
do not stand to the behavior that expresses them in such a way
Against the Self-Images of the Age
that there is any conceptual connection between the characteriza-
tion of the emotion and that of the behavior. I now want to develop
a second key thesis to the effect that statements about their own
behavior made by agents are not expressions of that behavior
at all because statements about their own behavior made by
agents stand in no different relationship to that behavior than do
statements made by others about their emotions. I thus have to
attack directly the Wittgensteinian view that first-person reports
are-just because they are expressions of feeling-not symmetrical
with second- and third-person ascriptions.
Against this I want to contend that there is no asymmetry, so
far as emotions are concerned (and if I restrict the point to emotions
in this essay this must not be taken to imply that I hold different
views about sensations), between first-person sentences on the one
hand and second- and third-person sentences on the other or
between statements expressed by means of first-person sentences
and statements expressed by means of second- and third-person
sentences. I take it that to understand personal pronouns at least
two conditions must be satisfied: first no one understands personal
pronouns who does not understand that they are blanks for which
personal proper names may be substituted. I do not understand
personal pronouns unless I am able to make the inference from
hearing someone say "MacIntyre is drunk" to "He is saying that
I am drunk." To have understood this is to be able to see what is
wrong with attempts to suggest that for "I" what can be sub-
stituted is not a proper name, but a description such as "the
present speaker." Of course, first-person statements can often be
correctly paraphrased by statements using such expressions as
"the present speaker." But "the present speaker" is not neces-
sarily self-referential in the way that "I" is. While I am speaking,
someone may say "The present speaker is drunk" and he would not
normally be taken to mean that he is drunk, but that I am; while
if I hear someone say "MacIntyre is drunk," I must if I under-
stand him see that it follows that if what he says is true then I am
drunk, but I do not need to have even learned the use of the
expression "the present speaker" to understand "I" and to use it
correctly.
Secondly, it is a condition of my understanding personal pro-
nouns that I understand that if I say truly of you that "You are
drunk," then you are able to say truly of yourself (provided that you
are not too drunk-perhaps this is what "can in principle" means)
"I am drunk"; and I can say truly of you to a third person, "He is
drunk" ; and in addition that if you can say truly of me that "You
Emotion, behavior and belief 239
are drunk," then I can say truly of myself "I am drunk"; and so on.
In other words, to have extricated oneself from the egocentric
predicament-insofar as this is a matter of grammar rather than
of beliefs-is a necessary condition for the exercise of the ability
to use those parts of speech which are held by some philosophers
to generate it.
It is these facts about personal pronouns which make it clear
that if the sentence "MacIntyre is angry" can be used truly to
make a statement about me, then "I am angry" can be used by me
to make the same true statement. This statement is, let us say, on
this particular occasion true, but it could have been false. The
statement made by my saying "I am angry" truly is thus a state-
ment which stands in the same relationship to my anger that the
statement "MacIntyre is angry" said by someone else stands to it.
But if this is so, then the statement cannot be an expression of my
anger, let alone the sentence. Certainly I may so utter the sentence
"I am angry" with clenched lips, gnashing teeth, and all the con-
ventional accompaniments of anger, whatever they are, in such a
way that my utterance is an expression of my anger. But my
utterance of "E =MC 2 " can be an expression of my anger in just
the same way as my utterance of "I am angry."
It is not to the point that I cannot be mistaken about whether
MacIntyre is angry if I am MacIntyre. I cannot be mistaken
because I cannot fail to have all the necessary evidence for what
I assert. But whenever I have all the necessary evidence for what
I assert, I cannot be mistaken either. Placed with vision unimpaired
in front of a tray containing a bottle and two glasses, I cannot be
mistaken in asserting that here are a bottle and two glasses. Yet of
course in both cases although I cannot be mistaken as to what is
true and what is false I can of course knowingly assert what is
false. Thus Ziff's acknowledgement that sometimes at least I can
only know whether you are angry if you will tell me implies that
I cannot know whether you are angry or not unless I know that
you are a trustworthy reporter of your emotions. But can I ever
know this?
It is not just that a great deal of behavior does not bear its
meaning on its face and that we cannot tell what intentions inform
it simply by observing it, because it is susceptible of more than
one and perhaps of many interpretations. But behavior which is
qua behavior unambiguous can be put to the service of a pretence,
just as assertions can be lies. This is the point at which to note
that when behaviorists have assimilated what I say to my behavior,
using such expressions as "linguistic behavior" or "verbal
240 Against the Self-Images of the Age
behavior" they have been wrong not so much in that they assimi-
lated them as in that they tried to reduce uses of language to forms
of behavior instead of seeing that certain forms of behavior are
best understood as at least resembling uses of language. One
reason why one of the key positions of this essay has not been
accepted by some philosophers-namely, that any behavior at all
can be expressive of certain emotions at least-is that there is a
behavioral iconography of emotion. That is, there are certain
forms of behavior which are by convention understood to sym-
bolize the presence of a particular emotion. To exhibit such be-
havior is equivalent to saying "I am angry" (or whatever emotion
is in question). It is because of this symbolic character of such
behavior-of shaking the fist and clenching the teeth in relation
to anger, for example-that it can be put to different uses just as
speech can. In Japan and Korea there is a tradition of visiting a
recently bereaved person who will then laugh and joke with his
guests without making reference to the bereavement. The host by
doing this asserts to his guests that he does not wish to burden
them with his grief; the guests assert in reply that they would not
wish to burden the host with the belief that he has in fact burdened
them with his grief. It is the conventional, symbolic character of
the behavior that makes this possible. So far as this iconographic
aspect of behavior is concerned, the conventions could, of course,
be other than they are. We could express anger by touching our
toes or gratitude by performing cartwheels. If it is objected that
very often the behavior expressive of anger or gratitude is evoked
from us by some action, is not a voluntary, deliberate, or controlled
response, it must be pointed out that very often too what we say
about our emotions is similarly evoked. An involuntary utterance
of a statement about my behayior does not, because it is involun-
tary, lose its conventional character. So too with behavior sym-
bolic of emotion.
There is, of course, a difference between speech and such
symbolic behavior which ought to be noted. I cannot use such
behavior to indicate to you that I am angry without being taken to
have expressed and not merely indicated the fact of my anger.
But this difference does not affect the preceding argument. What
follows from that argument is that whether I say to you that I am
angry or show you that I am angry I may be misleading you,
and from this, two additional consequences follow. The first is
that over large areas I can sometimes have no way of knowing
what you feel. Any performance of behavior by you, no matter
how extended, may be a pretence at the service of some further
Emotion, behavior and belief
unrevealed intention and emotion. This gap between performed
behavior on the one hand and intention and emotion on the other
is what goes unnoticed in a sociological perspective such as
Erving Goffmann's, in which there is nothing to human beings
but their performance of the behavior appropriate to different
roles and the behavior (equally and in precisely the same way
rule-governed) necessary to produce the behavior appropriate to
the roles. I introduce the doctrines of Goffmann in The Presenta-
tion of Self in Everyday Life at this point because in identifying
what is omitted from an account of social stituations which is
restricted to overt performances I have identified not merely what
is omitted from this particular piece of sociological inquiry and
description, but what it is at once essential and very difficult to
include. The early behaviorists believed that their doctrine pro-
vided the only basis for a scientific knowledge of human beings
and believed that such knowledge could have a firmer basis than
our everyday knowledge of each other has. Goffmann resembles
them in believing that the notion of "a true self" behind the role-
and rule-governed performances is an antiscientific myth per-
petrated by those who, as he puts it, wish to keep part of the
human world "safe from sociology." But in liquidating the dis-
tinction between the self and its performances, Goffmann loses
sight of the way in which we can only take what others do seriously
if we trust what they say on certain key occasions.
It is not just that the possibility of deception and of being misled
are so large; it is also that our beliefs about others have to be
founded so largely on trust if the preceding arguments are correct.
I may argue on an inductive basis about other people's emotions
insofar as these are not exhibited in or deducible from their
behavior; but however well contrived my inductive learning
policies are, the evidence from which I argue will include at crucial
points what others have said; and my willingness to treat what they
have said as trustworthy or untrustworthy cannot itself be induc-
tively based. Why not? Could I not learn that Smith is generally
trustworthy about his inner life too? The problem here is that we
could only derive well-founded inductive generalizations about
the connection between a man's trustworthiness in his monetary
transactions, say, and his trustworthiness in reporting his emotions
if we already had independent access to his emotions-which we
do not. Every piece of behavior is open to doubt. Does this entail
skepticism about the emotions of others? Or to compare the not
quite ridiculous with the not completely sublime, does my con-
clusion about the emotions of other people resemble Kant's
Against the Self-Images of the Age
conclusion about God: that the removal of knowledge has left room
for faith?
To put matters like this would be unhelpful; for clearly faith
in God is something that is dispensable. It is indeed of the essence
of faith that one can fail to have it. But if our beliefs about the
other people depend upon a presumption of their trustworthiness,
so that there is indeed a moral element in our beliefs about others
-and those who have wanted to mark a difference in this respect
between our knowledge of nature and our knowledge of others
have thus been right-it is not the case that we can rationally
decide not to make this presumption. For emotions are not
occurrences in the lives of individuals, insulated from similar
occurrences in the lives of other individuals. As Hume points out
in his discussion of sympathy in the Treatise, what I feel is in
large part a response to what I take others to feel or not feel. You
are resentful of my lack of gratitude at your generosity in the
face of my anger at your lack of sympathy for my depression over
your sentimentality. Such chains of emotion are characteristic
of the emotional life ; the plot of a novel often traces just such a
chain. It follows that systematic skeptical doubt about the emo-
tions of others, based on an acknowledgement of the opaque
quality of their behavior and a refusal to trust their avowals,
would produce an inability to respond to others, for we would
not know to what to respond and the reliefs that inform our
emotions would not specify adequate intentional objects for our
emotions. We have to trust one another at this basic level or be
paralysed in our humanity. This is not a choice.
There is one final point worth attending to. Small children
exhibit certain emotions spontaneously before they have learned
to pretend and about their emotions animals never learn to pre-
tend. So in these cases the element of trust in avowals is obviously
absent. Rage is an example of this. Small children also sometimes
learn in the case of certain other emotions to exhibit behavior
symbolically expressive of the emotion before they have learned
to feel the emotion. This is often true in the case of gratitude.
But what they have to learn in order to exhibit adult emotions
involves them in learning how to pretend, how to be ironic, how
to lie, and how to produce these stock responses which sustain
fatigued human relationships. In so doing they become, like the
adult world, opaque. Behaviorists were in the right when they
stressed that sometimes we want to claim that we can recognize
what others are feeling better than they can; but in repudiating
what introspectionists seemed to imply-that we are all continually
Emotion, behaviorand belief 2 43
opaque to each other all the time-they may have underrated the
extent to which we are very often opaque to each other a great
deal of the time. Misunderstanding and not understanding are
at the core of human life, a fact perhaps standing in the way of the
project of scientific inquiry about human beings, even if we view
it as an obstacle to be circumvented rather than as a final barrier.
This conclusion I find disconcerting. But perhaps the nature of
reality is such that we ought to have learned by now never to be
disconcerted at being disconcerted.
21
I
It is distinctly unfashionable to attempt to characterize and to
understand the history of human societies in terms of an opposition
between reason and irrationality. There was a time-in the closing
decades of the last century-when to do so was in fashion. Anthro-
pologists such as Frazer and Tylor, historians of thought such as
Lecky and Dickson White all took the possibility of so doing for
granted ; and it is of course in part against their parochialism and
against that misdescription of other cultures and other ages which
resulted from too great a confidence in the categories of their own
age that we have been reacting. Or overreacting, as I shall argue in
this essay. It is not that I want to return to the concepts of rationa-
lity and irrationality of the late Victorian Age; it is rather that I
do not want to see the perception of those concepts as culture-
bound to lead to a blindness to the importance of ascriptions of
rationality and irrationality in the human sciences.
The contemporary attitude is strikingly exemplified in the
strictures passed by H. R. Trevor-Roper 1 on the nineteenth-cen-
tury liberal historians of witchcraft, strictures with which it would
be difficult to quarrel. None the less, even if with Trevor-Roper we
reject such notions as that intellectual history is "a direct contest
between reason and faith, reason and superstition," we may find
reasons for holding that, when we have conceded that, as Trevor-
Roper holds, "such a distinction between 'reason' and 'supersti-
tion' is difficult to maintain, the making of some such distinction is
unavoidable. Indeed, we may find those reasons in Trevor-Roper's
own essay. Trevor-Roper explains the European witch-craze of the
sixteenth and seventeenth centuries as the outcome of certain social
1 "The European Witch-Craze" in Religion, the Reformation and Social
Change (London; Macmillan, 1967).
244
Rationality and the explanation of action 245
strains and conflicts. He speaks of "a scapegoat for social frustra-
tion," of "the mythology of the witch-craze" as "the articulation of
social pressure" and of "social tension." I do not want to, I am of
course not competent to question the truth of his explanation; but I
do want to try to understand its logical structure, and the attempt
to do this will lead me to use terms which are not used by Trevor-
Roper himself. I hope therefore that I do not misrepresent. When
it is said that social pressures generate a belief, as Trevor-Roper
asserts that social tensions between the people of the mountains
and the people of the plains partly generated the sixteenth-century
belief in witches, I take it that something like the following is
meant. A certain social situation generates certain emotions, in this
case a certain kind of fear. Emotions are intentional; that is, they
presuppose beliefs and we cannot characterize the emotion except
in terms of the relevant object of belief. So an expression of
fear is the expression of a belief that something harmful or
dangerous is at hand. The belief may, of course, be false; and of
course the frightened person may both express his fear and yet
assert sincerely that he does know that there is really nothing to be
afraid of. But where an emotion is generated which has no adequate
intentional object, such an object will under certain circumstances
be supplied. Jews, witches, and Communists have all at certain
periods been available as such objects for fear; not the Jew, the
witch, or the Communist of course as he or she is, but as the social
stereotype has him or her. The form of explanation is thus as
follows: we explain the belief as brought into being by a need to
supply a rationale for the emotion; and we explain the emotion as
generated by the social structure.
It is noteworthy that, although Trevor-Roper scorns the use of
"superstition" by Lea and Lecky, he himself is quite prepared to
call the belief in witches "a mythology" and a "fantastic" mythology
at that, and perhaps this is not as marginal a matter as he himself
seems to think. For we have to ask whether the type of explanation
which he advances would be equally appropriate for the explana-
tion of any set of well-established beliefs, or whether we must not
distinguish a class of beliefs for which it would be appropriate
and a class for which it would be inappropriate. Consider, for
example, the beliefs of a deviant minority in the early seventeenth
century, beliefs whose incompatability with the socially established
doctrines of the dominant institutions led to some sporadic
persecution. I refer to the astronomers. If we ask why by 1630
most astronomers believed that Jupiter had satellites, the explana-
tion must begin not only from Galileo's observations, but from
Against the Self-Images of the Age
what had become accepted canons of observation and of argument.
If we ask why those canons were accepted, the only possible
answer will be a historical one. This history may reveal to us
certain conditions which appear to have been necessary prere-
quisites for the exercise of the rationality embodied in those
canons; but just because they will only be necessary conditions,
they will in no way provide an explanation of, for example, why
those canons rather than any others were adopted. Moreover,
what such a history would not and could not reveal to us would
be antecedent sufficient conditions, sufficient, that is, to bring
about the social practice of arguing in terms of those canons. My
reasons for asserting this can be put as follows. Once we have
asserted that the explanation of why men came to believe and
believed something or other is that their behavior and procedures
were governed by the appropriate rational criteria, we have
already in so doing asserted that they had made themselves in-
dependent of these psychological or social factors which on
occasion lead men to act or to believe regardless of where reason
points. If some antecedent condition, such as a form of social
structure or the prevalence of certain emotions, is sufficient to
produce a belief, irrespective of the reasoning appropriately to be
invoked, then explanation in terms of the procedures of rationality
is clearly out of place. So when Trevor-Roper cites the beliefs of
modern anti-Semitism, of medieval anti-Semitism, and of
McCarthyism as parallels to the sixteenth- and seventeenth-
century witch-craze, he points to, although he does not formulate
precisely, a generalization of the form "Whenever the social struc-
ture is of a certain kind, beliefs of a certain kind will be generated
independent of any rational support they may or may not have."
In cases where such a generalization holds, the belief in question
will clearly be generated if the antecedent sufficient conditions are
satisfied, whatever the state of the processes of argument and
deliberation appropriate to the rational formation of such a belief.
Where, however, a belief is intelligible only in terms of an ante-
cedent process of reasoning and could only be generated as the
outcome of such a process, it would follow that its being held
cannot be ascribed to antecedent sufficient conditions of the kind
invoked by Trevor-Roper to explain the witch-craze.
It thus appears that there is at least a preliminary case for
holding that not all beliefs are to be explained in the same way.
The form of explanation appropriate to rational beliefs seems to be
not the same as the form of explanation appropriate to irrational
beliefs. We can well understand at a semi-intuitive level why this
Rationality and the explanation of action 247
should be so. To characterize a belief as irrational is to charac-
terize the intellectual procedures and attitudes of those who hold
it. It is to say in effect-at least in the extreme case-that the
believer is invulnerable to rational argument. But to say this is
precisely to say that the belief is held as the outcome of ante-
cedent events or states of affairs which are quite independent of
any relevant process of appropriate deliberation. Thus, pace
Trevor-Roper, a crucial distinction between rationality and
irrationality or superstition must continue to be made and we
cannot simply remark, as he does, that we have now learned that
the mental structures of one age are not the same as the mental
structures of another. For the explanation of rational belief
terminates with an acccount of the appropriate intellectual norms
and procedures; the explanation of irrational belief must be in
terms of causal generalizations which connect antecedent con-
ditions specified in terms of social structures or psychological
states-or both-with the genesis of beliefs.
Suppose to this it were to be retorted that the explanation of
rational beliefs cannot terminate with an account of the relevant
norms and rule-governed procedures. For any social practice
which is informed by such a body of norms and procedures must
itself be explained by references to prior antecedent conditions.
This retort might be grounded in the conviction that every event
or state of affairs must have a cause. Now certainly one may be
able to identify many conditions necessary for the establishment
and maintenance of such social practices and more especially
conditions which favor or inhibit their institutionalization. But
sufficient conditions? What could these be? The notion of a
causal explanation for the genesis of an intellectual tradition is like
the notion of such an explanation for the genesis of a style of
painting. All attempts to give such explanations have foundered.
The most that any specification of antecedent conditions can give
us (as Antal unintentionally demonstrated in his work on Floren-
tine painting) is a set of necessary and predisposing conditions
which might be shown to make a given audience receptive to
works of art characterized as falling under some very general
description. But the style which in fact procures the reception
always does so by virtue of characteristics which escape such
descriptions. The specific characteristics of Florentine painting
which make it great painting can in no way be deduced from the
market situation of the Florentine painter. As with art, so with
astronomy or arithmetic.
Two points ought to be noted about the argument so far. The
9
Against the Self-Images of the Age
first is that, even if we accept it, Trevor-Roper's strictures upon
the liberal historians of thought remain justified to a very large
degree. This is because what is in question in his discussion is not
merely the propriety of ascribing rationality or irrationality to the
beliefs which the social historian and the sociologist study; there
is also the question of whether the liberal historian's conception
of rationality was coherent and adequate. In fact it was not, for
they tended to confuse the question of rationality with that of
truth. This confusion has often been repeated ;1 the importance
of not repeating it will emerge later in my argument. For the
moment I will simply point out that "true" and "false" are
predicated of what is believed, namely of statements, and the truth
or falsity of a statement is a matter quite independent of whether
that statement is believed by anyone at all. Rationality is predi-
cated of the attitudes, dispositions, and procedures of those who
believe; a man who uses the best canons available to him may
behave rationally in believing what is false, and a man who pays
no heed to the rules of evidence may behave irrationally in believing
what is true. The parochialism of the liberal historians of the late
Victorian Age lay in their treating as one of the criteria of rationality
assent to the truth of certain propositions which they them-
selves affirmed; false belief they took to be a symptom of irrationa-
lity. In taking this to be the case, they certainly believed what is
false and they were perhaps irrational in so doing. For what
entitles us to call the belief in witchcraft irrational is not its
falsity, but the fact of its incoherence with other beliefs and
criteria possessed by those who held it.
Secondly, I would like to emphasize that this preliminary thesis
-that the type of explanation appropriate to a rational belief is
different from that appropriate to an irrational belief-is to be
distinguished from the thesis that human actions cannot be
causally explained. I have advanced reasons for denying this
thesis elsewhere.2 But it is of course the case that the present thesis
is not without implications for the explanation of actions. What
some of these are will be considered in the third section of this
essay. Before I consider them, however, another issue must be
raised.
II
From the conclusion of my argument in the previous section-
that the explanation of the genesis and maintenance of irrational
beliefs must be of a different order from the explanation of the
genesis and maintenance of rational belief-it follows that the
sociologist cannot, in setting himself the task of such explanation,
abstain from judgment concerning the rationality or irrationality
of the beliefs which he studies. It follows that the "neutrality"
thesis expressed by Bryan Wilson for the sociology of religion
can only be held in a tempered version. Bryan Wilson has argued
that the sociologist of religion, so long as he studies religious
belief and the religious believer qua sociologist, may and ought to
remain neutral as to the truth and reasonableness of the beliefs
which are part of the object of his study. Qua man he will doubtless
be theist or atheist, Catholic, humanist, or Parsee. But this need
not affect his work as a sociologist. About this thesis I should want
to make two comments. The first is that, for reasons I have already
given, the question of the truth or falsity of the belief studied
is to some degree independent of the question of its rationality;
but, although this distinction must not be ignored, truth and
rationality are both conceptually and empirically related. For to
advance reasons is always to advance reasons for holding that a
belief is true or false; and rational procedures are in fact those
which yield us the only truths of which we can be assured. Thus,
to recognize a belief as rationally held is to lay oneself open to at
least the possibility of its truth. Hence, if the sociologist qua
sociologist must, in order to determine what form of explanation
is appropriate to the genesis and maintenance of the beliefs which
he is studying, pass a verdict upon the reasonableness or unreason-
ableness of the procedures of those who hold the belief, he may
find that he is at least partially committed in respect of truth or
falsity.
Secondly, we must not as a result of the present argument ela-
borate a picture of the sociologist as approaching his material with
his canons of rationality which he uses to sit in judgment upon the
irrationalities of those whom he studies. We must distinguish at
least two elements in rationality. The first element is a matter of
language. To understand what is said in a given culture, we must
learn to classify the forms of utterance, to distinguish assertions
from requests, wishes from commands, and so on. We shall not
250 Against the Self-Images of the Age
be able to do this except on the assumption that the laws of logic
are embodied in the linguistic practice of the community which
we are studying. If we cannot identify negation and such laws as
,....,1>=1'
,...., as embodied in this practice, I do not understand how
we can be confident in our identification of the speech acts of
assertion and denial. It follows that any notion of pre-logical
thought, if it is taken to be thought which displays no regard for
the laws of noncontradiction, for example, is extremely difficult to
understand. Perhaps Levy-Bruhl never did seriously mean any-
thing as extreme as this-although in his earlier writings he does
seem to. So far as this element in rationality is concerned then,
there is no question of us judging the rationality of alien cultures
in terms of our criteria. For the criteria are neither ours nor theirs,
but simply the criteria, and logic is the inquiry which formulates
them.
This has sometimes been recognized, however, in such a way
as to draw a sharp and quite misplaced contrast between this
element and another. For it is suggested that on matters of sub-
stance, rather than matters of logic, what is believed to be a good
reason for holding a certain belief in one culture will appear as
in no way a good reason for holding that belief in the context of
another culture. Hence, it appears that what is counted as rationa-
lity in one culture must be something quite different from what is
counted as rationality in another. Peter Winch has written that
"We start from the position that standards of rationality in
different societies do not always coincide." 1 In his view we thus
can speak of our criteria and of theirs, where criteria rationality
are concerned. But this view fails-quite apart from the point
about logic made above-for two reasons. First, no belief is
rational or irrational except relative to some other belief or beliefs.
Thus, to predicate rationality or irrationality of the complete set
of beliefs held in a given culture is always a mistake. To say that
a belief is rational is to talk about how it stands in relation to other
beliefs, given a background of yet further beliefs as to what
counts as a good reason for holding beliefs on a particular type of
subject matter in a given culture. Because this is so, we can only
ascribe rationality to others on the basis of their criteria of rationa-
lity. Or rather, the distinction between ours and thei.rsagain breaks
down. If we indict others for contradiction or incoherence, the
contradiction or incoherence is a feature of their beliefs, but the
standards of contradiction and incoherence must be the same for
1 "Understanding a Primitive Society," American Philosophical Quar-
III
If we discriminate the rational and the irrational in the way that
I have suggested in the second section of this essay, and if the
form of explanation appropriate to irrational beliefs differs from
that appropriate to rational beliefs, as I have suggested in the first
section, what follows so far as the explanation of action is con-
cerned?
Beliefs and actions are, after all, intimately related, since it is a
central feature of actions that they are expressive of beliefs ; and
this is not just a contingent fact about actions. An action is identi-
fiable as the action that it is only in terms of the agent's intention.
2 54 Against the Self-Images of the Age
An intention can only be specified in terms of a first-person
statement. The expressions used in formulating such a statement
(even if the agent does not himself formulate it explicitly) will
presuppose certain beliefs on the agent's part. An everyday in-
tention to spray my roses to kill the green-fly or an exceptional
intention to assassinate an archduke and liberate Bosnia alike pre-
suppose an extended web of beliefs, botanical in the one case,
political in the other. It is for this reason that it is possible to
predicate of actions characteristics which it is the province of logic
to consider. An action may be consistent or inconsistent with an
agent's other beliefs, and one action may be consistent or incon-
sistent with another in terms of the beliefs presupposed. As
Aristotle pointed out, an action may conclude a syllogistic argu-
ment in a way analogous to that in which the utterance of a state-
ment may. It follows that the sociologist or anthropologist will
not even have succeeded in identifying correctly the actions which
are the object of his study unless and until he has identified the
web of beliefs expressed in those actions.
It is important to underline this because there is a tendency,
perhaps at its strongest in contemporary political science, to
suppose that the object of study in the human sciences is behavior,
and "behavior" is an expression understood in these quarters in a
behaviorist sense. Haunted by the ghosts of philosophical con-
troversies about dualism, such theorists wish to analyse all
mental predicates in behavioral terms and "belief" is an important
candidate for such an analysis. But no such analysis can succeed in
the case of "belief" at least. For all such analyses must, as their
proponents allow, include reference to what they quaintly call
"linguistic behavior." More particularly, if we try to analyse
the notion of belief in behavioral terms, then to say that someone
believes that such-and-such is the case will have to be analysed
not only in terms of dispositions to do and to expect certain things,
but also in terms of dispositions to say certain things. What sort of
disposition to say will be involved? The answer must be a dis-
position to make certain assertions. But what is an assertion? It is
the utterance of a statement in such a way as to give a hearer or
reader to understand that the statement is believed by the speaker
or writer and is worthy of belief. Thus, the notion of belief has not
been analysed away into behavioral terms, for the notion of asser-
tion-which any analysis that sought to be convincing would have
to employ-can itself be understood only in terms of the notion of
belief. So the notion of belief turns out to be ineliminable, and the
contemporary project of a science of behavior is seen to invert the
Rationality and the explanation of action 255
proper relationship between belief and action. Actions must be
understood in terms of their character as expressions of belief;
beliefs are not simply patterns of behavior plus dispositions to
produce such patterns.
I have already noted that to say that rational belief cannot be
explained in causal terms is not to say or to imply that actions, even
the actions of a man who acts upon a rational belief in a rational
way, cannot be explained in causal terms. Indeed, as I have
argued elsewhere,1 to treat an agent's actions as the outcome of the
reasons which he possessed for acting in the way that he did is
precisely to point to one kind of cause as operative and to exclude
other possible causal explanations. The notion that an agent's
having a reason to do something may be the cause of his doing it
is necessary if we are to distinguish reasons which are genuinely
effective from mere rationalizations which are not. But although
actions can have causes (in the sense of sufficient and not merely
of necessary conditions), the close link between actions and beliefs
would suggest that the asymmetry between the explanation of
rational belief for which I have argued ought to entail some
asymmetry between the explanation of rational action and the
explanation of irrational action. That it does so is made clear if we
consider two distinct types of cases which lie at opposite ends of a
dimension on which the relation of belief and action can be
charted.
At one end of this spectrum there is the case where an agent acts
rationally on the basis of his beliefs ; at the other end there is the
case where the agent's beliefs only affect his actions in the most
minimal way. There are cultures where the occurrence of the
latter type of case is a characteristic feature of social life ; in parts
of Latin America, for example, belief in sacramental monogamous
marriage is part of the Catholicism of the inhabitants, but the
actual forms of their sexual unions rarely, if ever, conform to the
Catholic pattern. There are other cultures (Campbell's account of a
Greek highland village is one example) where the professed beliefs
of the agents actually inform the detail of their day to day social
life. The difference between these types of cases is in the first
instance a difference in what requires explanation. For actions
which accord with the beliefs of an agent stand in need of no
further explanation than do the beliefs themselves; actions which
do not so accord clearly do stand in need of an independent
explanation, and the gap between belief and action itself requires
to be explained. It follows that once again there is an asymmetry,
1 See pp. 215-17.
Against the, Self-Images of the Age
although a rather different one. Where actions do accord with
beliefs, the form of explanation will be one in which the whole
complex of belief and action is to be explained together; and when
the beliefs are rational, explanation will terminate with the account
of the norms involved. Where, on the other hand, the beliefs are
irrational or the actions do not accord with beliefs, explanation
will 'have to go beyond the delineation of the relevant norms; for
1
IV
There was once a man who aspired to be the author of the general
theory of holes. When asked "What kind of hole-holes dug by
children in the sand for amusement, holes dug by gardeners to
plant lettuce seedlings, tank traps, holes made by roadmakers ?"
he would reply indignantly that he wished for a general theory
that would explain all of these. He rejected ab initio the-as he saw
it-pathetically common-sense view that of the digging of dif-
ferent kinds of holes there are quite different kinds of explanations
to be given; why then he would ask do we have the concept of a
hole? Lacking the explanations to which he originally aspired, he
then fell to discovering statistically significant correlations; he
found for example that there is a correlation between the aggre-
gate hole-digging achievement of a society as measured, or at least
one day to be measured, by econometric techniques, and its
degree of technological development. The United States surpasses
both Paraguay and Upper Volta in hole-digging. He also discovered
that war accelerates hole-digging ; there are more holes in Vietnam
than there were. These observations, he would always insist,
were neutral and value-free. This man's achievement has passed
totally unnoticed except by me. Had he however turned his talents
to political science, had he concerned himself not with holes, but
with modernization, urbanization or violence, I find it difficult to
believe that he might not have achieved high office in the APSA.
I
The ultimate aim of this paper is constructive; the skepticism
which infects so much of my argument is a means and not an end.
I do not want to show that there cannot be a general science of
political action, but only to indicate certain obstacles that stand in
260
ls a scienceof comparativepolitics possible? 261
the way of the founding of such a science and to suggest that the
present practice of so-called political science is unlikely to over-
come these obstacles. In writing more specifically of comparative
political science I do not wish to suggest that there could be any
other sort of political science; this the APSA recognized when it
merged what was its section devoted to comparative politics into
the general body. It is with the claim to be using legitimate
comparativemethods which could enable us to advance and to test
genuine law-like cross-culturalgeneralizations that I shall initially
be concerned. I shall not be concerned to question the possibility
of genuine and relevant comparison and even of cross-cultural
comparison for other purposes: to exhibit the march of the
Weltgeist through history, for instance, or to draw moral lessons
about the respective benefits of barbarism and civilization. These
may or may not be reputable activities; I shall not argue for or
against them here. I shall be solely interested in the project of a
political science, of the formulation of cross-cultural, law-like
causal generalizations which may in turn be explained by theories,
as the generalizations of Boyle's Law and Dalton's Law are
explained by the kinetic theory of gases ; all that I say about the
problem of comparability must be understood in this particular
context. Moreover, my skepticism about any alleged parallel
between theorizing about politics and theorizing about gases will
not initially be founded on the consideration of the character of
human action in general. I shall not argue, for example, that
human actions cannot have causes, not just or even mainly because
I believe that this proposition is false, but because I believe that,
even if its falsity is agreed, we still have substantial grounds for
skepticism about comparative political science. My method of
proceeding in the first part of my argument will be as follows: I
shall examine in turn the claim to have formulated law-like
generalizations about political attitudes, about political institutions
and practices, and about the discharge of political functions. I
shall then in the second part of my argument suggest an alternative
strategy to that now customarily employed, although the change
in strategy will turn out to also involve a change in aim.
II
The study of political culture, of political attitudes, as it has been
developed, seems to rest upon the assumption that it is possible
to identify political attitudes independently of political institutions
262 Against the Self-Images of the Age
and practices. There are at least two reasons for thinking this
assumption false. The first derives from Wittgenstein, who
pointed out that we identify and define attitudes in terms of the
objects toward which they are directed, and not vice versa. Our
understanding of the concept of fear depends upon our under-
standing of the concepts of harm and danger and not vice versa.
Our understanding of the concept of an aesthetic attitude depends
upon our understanding of the concept of a work of art. It follows
that an ability to identify a set of attitudes in one culture as
political, and a set of attitudes in some second culture as political,
with a view to comparing them must depend upon our ~aving
already identified as political in both cultures a set of institutions
and practices toward which these attitudes are directed. In other
words, the ability to construct comparative generalizations about
attitudes depends on our already having solved the problem of
how to construct comparative generalizations about institutions
and practices. The notion of political culture is secondary to and
parasitic upon the notion of political practice.
It follows that a necessary condition of a comparative investiga-
tion of political cultures is that the argument about the compara-
bility of political institutions should have a certain outcome; but
this is only a necessary end not a sufficient condition. It is also
necessary if political attitudes are to be the subject of comparative
inquiry that other attitudes shall be susceptible of comparison of
a certain kind. I can explain what I mean by this by citing an
example from The Civic Culture (Chapter IV, pp. 102-5) where
Almond and Verba argue that Italians are less committed to
and identified with the actions of their government than are
Germans or Englishmen, offering as evidence the fact that the
Italian respondents, as compared with the English and German
respondents to their survey, placed such actions very low on a list
of items to which they had been asked to give a rank order in terms
of the amount of pride they took in them. At no point do Almond
and Verba pause to ask whether the concept of pride is the same
in the three different national cultures, that is, to ask whether the
different respondents had after all been asked the same question.
But in fact the concept of pride(" ... si sente piu' orgoglioso ... ")
in Italy is not the same as that pride in England. The notion of
taking pride in Italian culture is still inexorably linked, especially
in the South but also in the North, to the notion of honor. What
one takes pride in is what touches on one's honor. If asked
to list the subjects which touched their honor, many Italians would
spontaneously place the chastity of their immediate female
Is a science of comparative politics possible? 263
relatives high on the list-a connection that it would occur to very
few Englishmen to make. These notions of pride and honor
partially specify and are partially specified by a notion of the
family itself importantly, if imperfectly, embodied in the actualities
of Italian family life. Hence we cannot hope to compare an Italian's
attitude to his government's acts with an Englishman's in respect
of the pride each takes; any comparison would have to begin from
the different range of virtues and emotions incorporated in the
different social institutions. Once again the project of comparing
attitudes independently of institutions and practices encounters
difficulties. These particular difficulties suggest that a key question
is: what are the units in each culture which are compared to be?
To this question I shall of course return; but let me note that the
difficulty which I have exemplified in the preceding argument
is contingent on Almond and Verba's particular procedures. It
does not arise from the project of comparison as such. For the
difficulty which arises over any comparison between English and
German culture on the one hand, and Italian on the other, from
relying on the in fact false assumption that these cultures agree in
their concept of pride would not arise in the same way if Italian
attitudes were to be compared with Greek, for example. Not that
there would not be other and perhaps more subtle pitfalls, but
these would not arise merely because concepts of pride and
honor are not shared.
We can now pose our problem in the following way: we wish
to find identifiable units in different societies and cultures about
which we may construct true causal generalizations. Political
attitudes, for the two reasons I have given, are implausible can-
didates; what about political institutions and practices? The first
point to be made here is that in turning to the discussion of
political institutions and practices we have not left behind the
topic of political attitudes. For attitudes to and beliefs about
institutions and practices may sometimes be purely external
phenomena; that is, the institution or the practice is what it is
and does what it does independently of what certain people think
and feel about it. But it is an obvious truism that no institution
or practice is what it is, or does what it does, independently of
what anyone whatsoever thinks or feels about it. For institutions
and practices are always partially, even if to differing degrees,
constituted by what certain people think and feel about them.
Consider the example of a currency system: a given type of
piece of paper or of metal has the value that it has not only because
it has been issued by a duly constituted authority, but because it is
Against the Self-Images of the Age
accepted as having that value by the members of a particular
currency-using population. When this condition is not generally
satisfied, as in Germany and Austria in 1923, the currency ceases
to have value, and thus ceases to be currency. So also with an
army: an officer has the authority that he has not only because his
commission has been issued by a duly constituted authority, but
because he is accepted as having that status by the men serving
under him. When this condition is not generally satisfied, as in
Russia in 1917, an officer ceases to have authority, and thus ceases
to be an officer. Since such beliefs about social institutions are
partially constitutive of social institutions, it is impossible to
identify the institution except in terms of the beliefs of those who
engage in its practices. This fact is ignored in general by those who
wish to define political science as the study of political behavior,
with a view to thereby providing a public, neutral subject matter
for scientific enquiry. But if we identify behavior except in terms
of the intentions and therefore of the beliefs of the agents we shall
risk describing what they are doing as what we would be doing if
we went through that series of movements or something like it
rather than what they are actually doing. Nor do we avoid this
difficulty merely by finding some description of the behavior in
question which both the agents themselves and the political
scientist would accept. For clearly both agents and political scientist
might apply the description "voting behavior" to what they do,
but yet have a quite different understanding of what it is to vote.
But now what bearing does all this have upon the project of
comparing political institutions and practices?
III
I take it that if the generalizations which political scientists con-
struct are to be part of a science, then among the conditions which
must be satisfied is this: that we shall be able to distinguish
between genuine law-like generalizations and mere def acto generali-
zations which hold only of the instances so far observed. I under-
stand by this distinction, as many others have understood by it,
the difference between a generalization the assertion of which
commits one to the assertion of a set of corresponding counter-
factual conditionals and a generalization which does not so com-
mit one. In the natural sciences the ground for treating a generaliza-
tion as a law is generally not merely that as a matter of fact no
plausible counter-examples have yet been produced. It is also nor-
Is a science of comparative politics possible? 265
mally required that it be supported by a body of theory. But what
then of these generalizations which we wish to assert as genuine law-
like generalizations before we have any well-established theory?
What about the generalizations of Kepler or of Galileo before
Newton formulated his laws? What about Boyle's Law or Dalton's
Law before the establishment of the kinetic theory? At this point
the problems of confirmation theory become real.
The particular finding of confirmation theory that is relevant is
that the degree to which a positive instance does genuinely confirm
a generalization is in part a matter of the kind of environment in
which it is found. For the greater the extent of the radically
different environments in which confirmatory instances of a genera-
lization are found, the less likely it is that the generalization is
only confirmed in certain contingent environmental circumstances.
Now it is a matter of contingent fact that nature is so structured
that this condition is normally realizable. For nature could have
been otherwise. If black ravens on being taken into laboratories
for pigmentation tests, or if black ravens on being observed in the
Arctic-in the course of our seeking confirmation or otherwise
of the generalization that all ravens are black-promptly turned
into philosphers of science or clouds of dust, generalizations about
ravenly nigritude could not be as well founded as they are. But in
fact the character of social life is such that in some respects it
resembles this imaginary nature rather than nature as it-for-
tunately for natural scientists-is.
Consider for example the alleged generalization that in two-
party electoral systems the two parties will tend to move together
in their policies and the alleged explanation for this generalization,
that this is because neither party can hope to win those voters
attracted by the furthest opposed wing of the other party, but only
those nearest to it. Hence where, for example, the parties and their
wings can be placed on a Left-Right dimension, each party tends
to move its policies toward the center, having no hope of winning
votes from the extreme Right or Left. Now consider two different
kinds of attempts to provide counter-examples to this generaliza-
tion. An example of the first would be Greece before the coup
d'etat of the colonels. This seems to be a straightforward refuta-
tion of the generalization, even if we remember that a single
counter-example in the natural sciences is never adequate to
refute a well-established theory or a generalization with a huge
weight of evidence supporting it, such as the generalization that
all solids except bismuth, cast-iron, ice, and type metal expand
when heated. For here we have nothing like a well-supported
266 Against the Self-Images of the Age
theory or generalization; it is rather as if the seventh raven we
were to come across was colored magenta. Now consider a quite
different kind of attempt to provide a counter-example.
Suppose that someone were to point to the rival parties in
Sierra Leone immediately before the army seized power there,
and to offer them as a counter-example. We ought at once to
remember what Ruth Schachter wrote of African mass parties:
"They and their cultural affiliates were interested in everything
from the cradle to the grave-in birth, initiation, religion, marriage,
divorce, dancing, song, plays, feuds, debts, land, migration, death,
public order-and not only electoral success." At once the question
cannot but be framed: "Why do we think of these as parties, rather
than as, say, churches?" The answer, that they have some of the
marks of American political parties, and that they call themselves
parties, does nothing to show that in fact the meaning of "party"
is not radically changed when the cultural context is radically
changed, or that even if it is not changed the: description has not
become inapplicable. The intentions, the beliefs, the concepts
which inform the practices of African mass parties provide so
different a context that there can be no question of transporting
the phenomena of party to this context in order to provide a
suitably different environment for testing our generalization.
Where the environment and where the culture is radically dif-
ferent the phenomenon is viewed so differently by those who
participate in it that it is an entirely different phenomenon. In
just this respect does society differ from nature. That is to say,
the provision of an environment sufficiently different to make the
search for counter-examples interesting will normally be the
provision of an environment where we cannot hope or expect to
find examples of the original phenomenon and therefore cannot
hope to find counter-examples.
Note that my thesis is not that to transplant a phenomenon such
as party is to subject it to causal influences which transform it.
That is doubtless true. But the difficulty of studying political
parties in alien social environments to test a generalization con -
structed about political parties in familiar social environments is
not like the difficulty of studying viruses : that their own causal
properties and/or those of the environment cause them to mutate
too rapidly and too often. If this were the type of difficulty that we
encountered in formulating cross-cultural generalizations about
politics, then we might well ask if we could not insulate the object
of study in its new environment from the disturbing causal
influences at work. To ask this would be to mistake my point
ls a scienceof comparativepolitics possible? 267
which is not about causal interference with the phenomenon of
party, but with the absence of the same concept of party, or perhaps
of any concept of party, as we understand it, in the alien culture.
Let me now consider a possible objection to this thesis which
would base itself upon my choice of examples. A quite different
choice of examples might provide us with more plausible candidates
for cross-cultural generalization. Consider the alleged (and quite
possibly false) generalization that in the government of cities, if
a single non-transferable vote for single members is the method of
election, then there will be over a certain time span a tendency
for a two-party system to flourish. This seems to hold in the
United States. But it might hold in other alien environments, even
environments of an exotic kind, where we could identify the system
as two-party, even if unclear in what sense the parties were parties.
But this is surely therefore an example of at least a possible cross-
cultural comparison which provides us with a law-like generaliza-
tion and is therefore lethal to my entire thesis. Let me at once
concede that I take this generalization to be law-like in that it
does indeed entail counter-factual conditionals, and let me further
concede that the counter-factuals in question might be true. But
I do not concede that it injures my thesis. Why not?
The reason for not conceding that this example, if true, would
injure my thesis is intimately connected with the fact that I should
not be extremely surprised if the generalization in question did
turn out to be true of cities outside North America as well as in
North America. For what could make the generalization true, if
true, is that voters prefer in general not to waste their votes in
voting on matters that concern the administration of their daily
lives; and it requires only a minimal and a very untheoretical
understanding of the electoral system produced by such a voting
procedure to understand that in the majority of cases votes for a
third party will be wasted. The considerations from which we
can deduce this particular generalization are thus concerned with
human rationality in general; they do not have any specific con-
nection with politics and they do not belong to political science,
but to our general understanding of rationality. This will be true
of all generalizations which concern the formal structures of
human argument, even if they appear in political clothing, furnish-
ing us with explanations of particular political choices and actions.
So it must be, for example, with all applications of the theory of
games to politics.
My thesis about the legitimacy or otherwise of the project of
accumulating a stock of cross-cultural generalizations about
Against the Self-Images of the Age
political behavior to furnish the empirical foundation for a political
science, as I have developed it so far, can now be stated disjunc-
tively: either such generalizations about institutions will neces-
sarily lack the kind of confirmation they require or they will be
consequences of true generalizations about human rationality
and not part of a specifically political science.
To complete this part of my argument I must now make three
further observations. The first is that my statement of the diffi-
culties in constructing true and warranted cross-cultural generali-
zations about political institutions is obviously akin to the argu-
ments which some anthropologists-notably Edmund Leach and
Walter Goldschmidt-have developed about cross-cultural gene-
ralizations in their discipline. But Goldschmidt has then argued
that it is not institutions, but functions, or rather institutions only
as serving certain functions, which we ought to aspire to compare;
and this contention has already been advanced by some political
scientists. We are, that is to say, to begin by identifying the same
function in different societies and then to inquire how quite
different institutions have this same effect; for I take it that to say
that X performs, serves, or discharges a given function always
entails that X is the cause of a particular effect, even if this does
not exhaust the meaning of the statement in which function was
ascribed. It is certainly not a final objection to this project that
most political scientists who have tried to specify the functions in
question have produced nothing but statements about institutions
and their effects in which the word "function" may appear, but
could be replaced not only without loss, but with gain. "Wherever
the political party has emerged, it appears to perform some com-
mon functions in a wide variety of political systems ... the organi-
zation called the party is expected to organize public opinion and
to communicate demands to the center of governmental power and
decision . . . the party must articulate to its followers the concept
and meaning of the broader community ... the party is likely to
be involved in political recruitment . . . These similarities of
function ... suggest that the political party when the activities
of a political system reach a certain degree of complexity, or
whenever the notion of political power comes to include the idea
that the mass public must participate or be controlled. " 1 In a
passage like this, the notion of function can be replaced entirely
by either the notion of effect or the notion of purpose. When
we so replace it, we notice also that the transition from previous
1 J. LaPalombara and M. Weiner, eds., Political Parties and Political
Development (Princeton, N.J.: Princeton University Press).
ls a science of comparative politics possible? 269
to tentative conclusion requires no reliance on any factual genera-
lizations anyway; it is merely a matter of drawing out the con-
sequences of definition. But even if in the writing of political
scientists as sophisticated as LaPalombara and Weiner the
function of the use of "function" is unclear, it does not follow
that this has to be so. But the condition of its not being so is that
we should have some criteria for identifying the functions served
by political institutions which is other than, and independent of,
the aims and purposes of political agents and the effects of political
institutions. The provision of such a criteria would require the
identification of a system, using the word "system" precisely, so
that concepts of feedback and equilibrium are applicable on the
basis of quantitative data which will provide values for variables
in differential equations. I scarcely need stress the remoteness
of this goal from the present state of all political science; if we
match the requirements that have to be satisfied to identify
such a system-which would involve, for example, being able
to distinguish between change that is part of the movement of
items through the system, change that is itself part of the struc-
turing of the system, and change that is the system decaying by
providing ways of measuring rates of change for all three-then
a work like David Easton's A Systems Analysis of Political Life
looks like a mad, millenarian dream. I therefore take it that any
attempt to answer my argument by suggesting that cross-cultural
generalizations about institutions may be provided by means of a
prior account in terms of functions is bound to fail.
My second observation is that my argument does not imply any
undervaluation of the importance of the work done by political
scientists in establishing both the facts about particular institutions
and the very limited generalizations they do establish. That the
conditions under which these generalizations hold necessarily
remain unclear to us for the kind of reason that I have given
does not mean that we do not need the best that we can get in this
case, which is what they give us; only this kind of accumulation
of data in no way leads toward the construction of a science. I
shall later suggest an alternative context in which these empirical
labors could perhaps be viewed more constructively. For the
moment I note that it is Machiavelli who ought to be regarded as
the patron saint of political studies and not Hobbes, and for this
reason: Hobbes believed-as presumably Almond and La-
Palombara and Easton (although Easton, in ways that I do not
entirely understand, tries to distinguish his enterprise from that of
Hobbes) believe-that the fortuitous, the surprising, the unpre-
Against the Self-Images of the Age
dieted, arise in politics only because our knowledge of political
motions is less adequate than our knowledge of planetary motions.
Given time, labor, and foundation grants-the contemporary
version of royal patronage--an unpredicted revolution-but for
the sheer complexity of human affairs-ought to be as disgraceful
to political scientists as an unpredicted eclipse to astronomers.
But Machiavelli realized that in political lifefortuna, the bitch god-
dess of unpredictability, has never been dethroned. To any
stock of maxims derived from empirically founded generalizations
the student of politics must always add one more: "And do not
be surprised if in the event things turn out otherwise." The need to
include this maxim follows from my argument, just as it follows
from Machiavelli's.
My third observation is that in the history of political theory we
have more than once been here before, and notably in the dispute
between James Mill and Macaulay. James Mill argued, although in
the interests of a quite different conclusion, even more that we
cannot find reliable empirical generalizations about political
behavior: "Absolute monarchy under Neros and Caligulas ... i~
the scourge of human nature. On the other side, the public of
Denmark ... under their absolute monarch are as well governed as
any people in Europe . . . the surface of history affords, there-
fore, no certain principles of decision." Mill then proceeded to
argue from this that we ought to turn instead to the type of psy-
chology favored by the utilitarians for our explanations, that there
is no specifically political science. Against him Macaulay argued
that the empirical facts about government do yield genuine law-
like generalizations, not least generalizations of a kind which
enable us to predict future actions with great confidence. And
it is clear that this practical use of law-like generalizations provides
Macaulay with a crucial motive. The claim to technical expertise
on the part of the political scientist is closely bound up with the
defense of the possibility of formulating law-like generalizations.
If the latter fails, the former is gravely impaired. When in our time
on the basis of his generalizations Lipset predicts totalitarian
horrors as the outcome of widespread political participation, he
turns out to be the true heir of Macaulay who, on the basis of his
generalizations, predicted cultural ruin if "the great number" were
allowed to participate in government; "they will commit waste of
every sort in the estate of mankind, and transmit it to posterity
impoverished and desolate," so that "in two or three hundred
years, a few lean and half naked fishermen may divide with owls
and foxes the ruins of the greatest of European cities ... " In
ls a scienceof comparativepolitics possible? 271
IV